Professional Documents
Culture Documents
ACTIVE ANTENNAS
ACTIVE ANTENNAS
The present article is an introduction to the topic of active
antennas. The first section is a description of the field suitable
for reading by almost any undergraduate science major. The
next section is an in-depth reexamination of the subject, including equations and some derivations. Its basic idea is to
provide the reader with enough tools that he or she can evaluate whether it is an active antenna that he or she might need
for a specific application. The final section is a discussion of
where active antennas are finding and will find application.
We should mention here that, if one really needs to design
active antennas, one will need to go further than this article.
The set of references to the primary research literature given
in this article is by no means complete, nor is it meant to be.
A good way to get started on the current literature on this
topic would be a reading of the overview monograph of Navarro and Chang (1). We will not cover active amplifiers in
this article. However, this topic is treated in the book edited
by York and Popovic (2).
AN INTRODUCTION TO ACTIVE ANTENNAS
An antenna is a structure that converts electromagnetic energy propagating in free space into voltage and current in an
electrical circuit and/or vice versa. In a transceiver system,
the antenna is used both to receive and to transmit free-space
waves. At minimum, a transceiver then must consist of a signal source that serves to drive the antenna as well as a receiver circuit that reads out the signal from the antenna. Until recently, practically all antenna systems operating in the
microwave frequency regime (operation frequencies greater
than 1 billion cycles per second, or 1 GHz) were mostly designed to isolate the antenna from the circuitsthat is, to
find ways to make system operation independent of the antennas electrical characteristics. In contradistinction, an active antenna is one in which the antenna actually serves as a
circuit element of either the driver or the readout circuit. To
understand why this is different from conventional antenna
driving or readout will require us to take a bit of a historical
trip through the last century or so.
Actually, the first antenna was an active one. Heinrich
Hertz, back in 1884 (2a), was the first to demonstrate that
one could generate radio waves and that they would propagate from a transmitter to a receiver at the speed of light.
The apparatus used is schematically depicted in Fig. 1. The
idea of the transmitter is that, by discharging an induction
coil (a wire looped about a magnetic core such that the composite device can store significant amounts of magnetic energy) into a spark gap, one can generate a current in the 5
mm diameter wire. The voltage in the spark gap induces a
current in the wires, which in turn induces a voltage in the
wires, and this voltage in turn induces current, so that the
voltage and current propagate along the two pieces of the wire
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
Induction
coil
5 mm
3/4-cm
spark gap
1m
35 cm
30 cm
2 mm
(a)
(b)
Figure 1. Hertz apparatus for (a) transmitting and (b) receiving radio waves, where the transmitting antenna serves to choose a specific
frequency of the spark gap voltage to transmit to the receiving antenna, which also serves to pick out this special frequency from the
free-space waveform and turn this electromagnetic disturbance into
a voltage across the receiver antenna gap.
to either side of the gap as waves, appearing much like a onedimensional slice through a water wave propagating away
from the point where a pebble has struck the waters surface
(the spark gap). A wave will propagate rectilinearly until it
encounters an obstruction, at which point it can suffer reflection from or transmission into the barrier that the obstruction
presents. There will be reflections then off the metal spheres on
the ends of the wire. The spark will generate a broad spectrum
of frequencies or wavelengths. The reflections off the two ends,
though, will tend to cancel each other except at certain special
frequencies. The effect at these wrong frequencies is much like
the effect of throwing a handful of pebbles into the pond and noting that, in between the points where the pebbles struck, the
waves are much more indistinct than they are far from where
the handful struck the surface. The special frequencies are ones
which just fit into the region between the spheres. The current
needs to be zero at the two ends in order to fit, whereas the voltage needs to be maximum at the ends. The current and voltage
waves at the right frequencies may appear as depicted in Fig. 2.
The Hertz transmitter is the archetypical active antenna.
The source is the spark gap, which is actually placed in the
antenna. The antenna then acts as a filter to pick the right
frequency out of a large number of frequencies that could be
launched from the gap. The receiver is picked to be of a length
to also select this primary frequency.
Hertz-style spark gap transmitters, after further development and popularization by Marconi, were in use for fifty
years after Hertz. However, such transmitters exhibit some
rather severe drawbacks. The main problem is that the simple resonant dipole antenna (that is, a straight wire antenna
with a gap or a feeder cable used to feed in current) is a filter
with a poor frequency selection. Namely, if one increases the
I
V
Power transmitted
ACTIVE ANTENNAS
191
100%
75%
f0
2f0
3f0
4f0
Figure 3. A sketch of what the transmission as a function of frequency might look like for the Hertzian dipole antenna of Figs. 1
and 2.
192
ACTIVE ANTENNAS
L
Battery
Band
Designation
Frequency
(GHz)
Wavelength
L
S
C
X
Ku
K
Ka
Q
U
V
E
W
D
G
Y
12
24
48
812
1218
1826
2640
3350
4060
5075
6080
75110
110170
140220
220325
1530 cm
7.515 cm
3.757.5 cm
2.53.75 cm
1.672.5 cm
1.151.67 cm
0.751.15 cm
69 mm
57.5 mm
46 mm
3.755 mm
2.74 mm
1.82.7 mm
1.42.1 mm
0.91.4 mm
Ground
Figure 5. A circuit with lumped elements connected by wire segments.
Area A
++++++++++++++
RS
d
RL
ACTIVE ANTENNAS
Inner
conductor
Outer conductor
(ground potential)
Figure 7. A coaxial cable in which signals are carried on an inner
conductor and in which the grounded outer conductor serves to carry
the ground plane along with the signal in order to give a constant
impedance along the line.
Oscillator
Antenna
Matching
network
Matching
network
Transmission
line
Transmission
line
193
Matching
network
Matching
network
Amplifier
Figure 8. Schematic of a conventional RF microwave transmitter in which each individual element of the transmitter is
matched to each other element.
194
ACTIVE ANTENNAS
Dc
power
in
Input 1
Input 2
Metal
Dielectric
Output
a
b
(a)
(b)
Figure 10. Views of (a) a microstrip and (b) a coplanar waveguide
line. In the microstrip, the ground plane is the lower electrode,
whereas in the coplanar waveguide the ground plane is placed on the
surface of the dielectric substrate.
Output strip
Gap
Drain
ACTIVE ANTENNAS
(a)
195
(b)
(a)
;
;
;;
(b)
Input line
(c)
Figure 13. Illustration of the electric field directions along (a) the
nonradiating edge and (b) the radiating edge, and (c) a schematic
depiction of the edge fields around the patch.
that is, the fields that point neither up nor down but rather
across. Beyond the longitudinal edges of the patch are fields,
in phase for the two edges, that are normal to the surface. It
is these fields (when combined with transverse magnetic
fringe fields in the same strips) that give rise to the upward
radiation. Similar arguments describe the operation of the slit
antenna if one exchanges the electric and magnetic fields in
the argument.
We have now introduced all of the pieces necessary to describe the new resurgence in active antenna research. A possible active antenna design could appear as in Fig. 14 (7),
where the transistor is actually mounted right into the patch
antenna element, and therefore the design can be quite compact. That is, the source plus oscillator plus antenna can all
be fitted into less than a wavelength. The design of Fig. 14,
which comes from R. Comptons group at Cornell (31,32), will
be discussed further in the next section.
/4
short
Gate
Source
Source
Slit
Drain
Drain
bias
Bypass
capacitor
Source leads,
grounded
Figure 14. Depiction of the upper surface metallization of a microstrip active patch antenna discussed in Ref. 7. The short on the gate
together with the slit between gate and drain provides the proper
feedback delay to cause oscillation.
196
ACTIVE ANTENNAS
There are a number of advantages to the use of active antennas. One is that an active antenna can be made compact.
Compactness in itself is advantageous, as throughout the history of microelectronics, miniaturization has led to lowered
costs. There are two more advantages, though, which relate
to compactness. One is that the power-handling capabilities
of a device go down with increasing frequency. We would
therefore like to find ways to combine the power from several
devices. One can try to add together outputs from various oscillators in the circuit before feeding them to the elements,
but this goes back to the conventional solution. A more advantageous design is to make an array of antennas, with proper
spacing relative to the wavelength and antenna sizes, and
add the power of the locked oscillators in the array quasioptically in free space. (In other words, optical radiation tends
to radiate into free space, whereas radio frequency in microwave radiation needs to be kept in guiding waveguides until
encroachment on radiating elements. Quasi-optics uses the
principle of the optical interferometer to combine multiple coherent microwave fields in free space.) The locking requires
that the oscillators talk to each other so that the phases of all
of the array elements stay in a given relation. As will be discussed in more detail in the next section, however, an important problem at present in the active antenna field relates to
keeping elements locked yet still being able to modulate the
output as well as steer the beam in order to be able to electronically determine on output direction. These issues will be
discussed in the next section and taken up in more detail in
the last section.
where E is the electric field vector, B is the magnetic induction vector, H is the magnetic field vector, D is the electric
displacement vector, J is the current density vector, and is
the volume density of charge. An additional important quantity is S, the Poynting vector, defined by
B
B
t
D
D
H = J +
t
D =
E =
B = 0
S = E H
If one takes the divergence of S, one finds
E H)
S = (E
If one assumes a free-space region,
D = 0E
B = 0 H
which is therefore lossless,
J =0
and charge-free,
=0
(where 0 is the permittivity of free space and 0 is the permeability of free space), one can use vector identities and Maxwells equations to obtain
S =
0
E E ) 0 (H
H H)
(E
2 t
2 t
where dA is the differential area times the unit normal pointing out of the surface of the volume V, one finds that
S dA =
We Wm
t
t
0
E E dV
2
0
H H dV
2
The interpretation of the above is that the amount of S flowing out of V is the amount of change of the energy within.
One therefore associates energy flow with S E H. This is
important in describing energy flow in wires as well as transmission lines and waveguides of all types. As was first described by Heaviside (9), the energy flow in a wire occurs not
inside the wire but around it. That is, as the wire is highly
conductive, there is essentially no field inside it except at the
surface, where the outer layer of oscillating charges have no
outer shell to cancel their effect. There is therefore a radial
electric field emanating from the surface of the wire, which
ACTIVE ANTENNAS
combines with an azimuthal magnetic field that rings the current flow to yield an E H surrounding the wire and pointing
down its axis.
It was Pocklington in 1897 (10) who made the formal structure of the fields around a wire a bit more explicit and, in the
effort, also formed the basis for the approximation upon
which most of circuit and transmission line theory rests, the
quasi-static approximation. A simplified version of his argument is as follows. Assume an xyz Cartesian coordinate
system where the axis of the wire is the z axis. One then
assumes that all of the field quantities f(x, y, z, t) vary as
f (x, y, z, t) = f (x, y) cos(z t + )
If one assumes that the velocity of propagation of the abovedefined wave is c (00)1/2, the speed of light, then one can
write that
=
+ e y
t = e x
x
y
which is just the transverse, and therefore two-dimensional,
gradient operator. These equations are just the electro- and
magnetostatic equations for the transverse fields, whereas
the propagation equation above shows that these static transverse field configurations are propagated forward as if they
corresponded to a plane wave field configuration. If the magnetic field is caused by the current in the wire, it rings the
wire, whereas if the electric field is static, it must appear to
emanate from charges in the wire and point outward at right
angles to the magnetic field. If this is true, then the Poynting
vector S will point along the direction of propagation and the
theory is self-consistent, if approximate.
If we wish to guide power, then the quasistatic picture
must come close to holding, as the Poynting vector is in the
right direction for guidance. The more general approximate
theory that comes from Pocklingtons quasistatic approximation is generally called transmission line theory. To derive this
theory, first consider the two-wire transmission line of Fig.
197
Ground plane
Equiphase
lines
C1
z axis
J lines
V2
Conductor 1
E dA 1 =
B
B
dA 1
t
(Li)
t
198
ACTIVE ANTENNAS
i(z)
v(z)
l
vi
(1)
=0
t
J dV2 =
Q=
dV2
Q = Cv
Noting that
J dV2 =
J dA 2
J dA 2
vo
g vo
tation, as is schematically depicted in Fig. 16(a). One can verify this by writing Kirchhoff s laws for the nodes with v(z
dz) and v(z) using the relations
v=l
i
t
i=c
v
t
lc
=0
z2
t 2
which is a wave equation with solutions
and
Figure 16. A circuit equivalent for (a) a lossless and (b) a lossy transmission line. The actual stages should be infinitesimally long, and
the ls and cs can vary with distance down the line. In reality, one
can find closed-form solutions for the waves in nominally constant l
and c segments and put them together with boundary conditions.
dV2
t
(b)
vi
(a)
H) = J + D
(
t
v(z + dz)
l
(2)
(3)
;;;
;;;
ACTIVE ANTENNAS
l
i
c f
l
i
vb =
c b
vf =
= lc
Figure 17. Schematic depiction of a top view of the metallized surface of an FET, where G denotes gate, D drain, and S source.
vp =
l/c
Z(z ) =
Z + jZ0 tan (z l)
v(z l)
= Z0 l
i(z l)
Z0 + jZl tan (z l)
R=
i(z, t) = Re[i(z)e jt ]
so that we can write
v
= jli
z
i
= jcv
z
with solutions
v(z) = vf e j z + vb e j z
i(z) = if e j z ib e j z
Z0 i(l) = vf e j l vb e j l
hold, and from them we can find
vf =
vb =
Zl Z0
Zl + Z0
;;;;;;
;;;;
Zl i(l) = vf e j l + vb e j l
+ Z0 )i(l) e
j l
Z0 )i(l) e
j l
which gives
Schottky barrier
S
n+
Ohmic contact
i(l)
[(Zl + Z0 ) e j (lz) + (Zl Z0 ) e j (lz) ]
v(z) =
2
i(l)
[(Zl + Z0 ) e j (lz) (Zl Z0 ) e j (lz) ]
i(z) =
2Z0
(4)
This equation allows us to, in essence, move the load from the
plane l to any other plane. This transformation can be used
to eliminate line segments and thereby use circuits on them
directly. However, note that line lengths at least comparable
to a wavelength are necessary in order to significantly alter
the impedance. At the plane z l, then, we can further note
that the ratio of the reflected voltage coefficient vb and the
forward-going vf , which is the voltage reflection coefficient, is
given by
v(z, t) = Re[v(z)e jt ]
1
(Zl
2
1
2 (Zl
199
Ga
As
Ohmic contact
Figure 18. Schematic depiction of the cross section of the active region of a GaAs FET. Specific designs can vary significantly in the
field-effect family.
200
ACTIVE ANTENNAS
+Vds
Vgs = 2 V
G
Ids
5V
(a)
Rd
Cgs
vin ZT
V
g mV
Rds Cds
ZL vout
Vds
10 V
(b)
Figure 19. (a) Circuit element diagram with voltages and currents
labeled for (b), where a typical IV curve is depicted.
The next piece of an active antenna that needs to be discussed is the active element. Without too much loss of generality, we will take our device to be a field effect transistor
(FET). The FET as such was first described by Shockley in
1952 (5), but the MESFET (metalsemiconductor FET),
which is todays workhorse active device for microwave circuitry, was not realized until 1965 (6), when gallium arsenide
(GaAs) fabrication techniques became workable albeit only as
a laboratory demonstration. [Although we will discuss the
MESFET in this section, it should be pointed out that the
silicon MOSFET (metaloxidesemiconductor FET) is the
workhorse device of digital electronics and therefore the most
common of all electronic devices presently in existence by a
very large margin.] A top view of an FET might appear as in
Fig. 17. As is shown clearly in the figure, an FET is a threeterminal device with gate, drain, and source regions. A cross
section of the active region (that is, where the gate is very
narrow) might appear as in Fig. 18. The basic idea is that the
saturation-doped n region causes current to flow through the
ohmic contacts from drain to source (that is, electrons flow
from source to drain), but the current is controlled in magnitude by the electric field generated by the reverse bias voltage
applied to the gate electrode. The situation is described in a
bit more detail in Fig. 19, where bias voltages are defined and
a typical IV curve for dc operation is given. Typically the
bias is supplied by a circuit such as that of Fig. 20. In what
follows, we will simply assume that the biases are properly
applied and isolated, and we will consider the ac operation.
An ac circuit model is given in Fig. 21. If one uses the proper
number of circuit values, these models can be quite accurate,
but the values do vary from device to device, even when the
V1
S11 S12
V1+
=
(5)
V2
S21 S22
V2+
An important parameter of the circuit design is the transfer function of the transistor circuit, which can be defined as
the ratio of vo to vi as defined in Fig. 21. To simplify further
analysis, we will ignore the package parasitics Rg and Rd in
comparison with other circuit parameters, and thereby we
will carry out further analysis on the circuit depicted in Fig.
23. The circuit can be solved by writing a simultaneous system of equations for the two nodal voltages vi and vo. These
sinusoidal steady-state equations become
vi = v
jCgd (vo vi ) + gm vi + jCds vo +
vo
vo
+
=0
Rds
ZL
+Vds
T=
Vgs
Zt
Intrinsic ac model
Vgs = 3 V
vi
Cgd
+
Vgs = 1 V
D
Vgs
Rg
Vgs = 0
Idss
ZL
vo
=
vi
gm + jCgd
1
1
j (Cgd + Cgs ) +
+
Rds
ZL
vo
Figure 20. Typical FET circuit including the bias voltages vgs and
vds as well as the ac voltages vi and vo, where the conductors represent
ac blocks and the capacitors dc blocks.
V 1+ej z
V1 e j z
Two-port network,
biased FET
S
V 2+ e j z
V 2 e j z
ACTIVE ANTENNAS
ii
+
v ZT
vi
Cgs
io
vo
Cgd
RT
+
v
Cds
g mv
Rds
ZL
Zi
ZT
Ri
jXT
201
gm Rds + jCgd Rds
vo
=
vi oc
j (Cgd + Cgs ) Rds + 1
jXi
Figure 24. Diagram depicting the transistor and its load as a oneport device that, when matched to its termination so that there is
no real or imaginary part to the total circuit impedance, will allow
for oscillations.
1
Rds
Cds
vo
G( j)
=
vi
1 + G( j)H( j)
gm Rds
R
1+
ZL
G = gm Rds
T=
H( j ) =
Clearly, from the above, one sees that the loaded gain will be
lower than the unloaded gain, as we would expect. Making
only the first of our two above approximations, we can write
the above equations as
T=
gm Rds
1 + jds +
G=
Rds
ZL
gm Rds
1 + jds
Zi =
ii
= jCgs + jCgd
1 +
vi
1
Cds Rds
1
G( j)
gm jCgd
1
1
j (Cgd + Cds ) +
+
Rds
ZL
+
vi
G( j)
vo
H( j )
Figure 25. Depiction of a simple feedback network.
202
ACTIVE ANTENNAS
gm Rds + 1 + jds +
Zi = jCgs + jCgd
Rds
ZL
R
1 + jds + d
ZL
The condition for oscillation in such a system can be expressed in either of the forms
i T = 1
or
o L = 1
ZL (1 + gm Rds + Rds )
Rds + ZL
|T | < 1
|L | < 1
ZL = j L
and
gm Rds 1
|i | < 1
|o | < 1
then clearly
Ri < 0
Whether or not Xi can be made to match any termination is
another question, which we will take up in the next paragraph.
As was mentioned earlier, generally the data sheet one obtains with an FET has plots of the frequency dependence of
the S parameters rather than values for the equivalent circuit
parameters. Oscillator analysis is therefore usually carried
out using a model of the circuit such as that depicted in Fig.
26, where the transistor is represented by its measured S matrix. The S matrix is defined as the matrix of reflection and
transmission coefficients. That is to say, with referrence to
the figure, S11 would be the complex ratio of the field reflected
from the device divided by the field incident on the device.
S21 would be the field transmitted from the device divided by
the field incident on the device. S12 would be the field incident
from the load side of the device divided by the power incident
on the device, and S22 would be the power reflected from the
load side of the device divided by the power incident on the
device. For example, if there is only an input from ZT, then
xn =
n=0
1
1x
S12 S21 L
1 S22 L
S S
o = S22 + 12 21 T
1 S22 T
i = S11 +
i = S11
o = S22
=
S
ZT
T i
ZL
o L
ACTIVE ANTENNAS
i
1
+ (Ri + RT )i +
i dt = 0
t
C
vi
t
t 2
L
mv2
Ri RT
a
t
b
Dielectric
constant
x
v
v
+
=0
t
LC
v
2v
+ 02 v = 0
(1 2 v2 )
t 2
t
which is the form of Van der Pols equation (15,16), which
describes the behavior of essentially any oscillator.
Now that we have discussed planar circuits and dynamical
elements that we can put into theory, the time has arrived to
discuss planar antenna structures. Perhaps the best way to
gain understanding of the operation of a patch antenna is by
considering a cavity resonator model of one. A good review of
microstrip antennas is given in Carver and Mink (17) and is
reprinted in Pozar and Schaubert (18). Let us consider a
patch antenna and coordinate system as is illustrated in Fig.
28. The basic idea behind the cavity model is to consider the
region between the patch and ground plane as a resonator. To
do this, we need to apply some moderately crude approximate
boundary conditions. We will assume that there is only a zdirected electric field underneath the patch and that this field
achieves maxima on the edges (open-circuit boundary condition). The magnetic field H will be assumed to have both x
and y components, and its tangential components on the
edges will be zero. (This boundary condition is the one consistent with the open-circuit condition on the electric field and
becomes exact as the thickness of the layer approaches zero,
as there can be no component of current normal to the edge
at the edge, and it is the normal component of the current
that generates the transverse H field.) The electric field satisfying the open-circuit condition can be seen to be given by the
modes
mn
cos kn x cos km y
e mn = e z
abt
ii
where
+
RT
Vi
203
Ri
kn = n/a
L
km = m/b
m=0
1,
mn =
2, m = 0
2,
m = 0
and n = 0
or n = 0
and n = 0
204
ACTIVE ANTENNAS
b
vi
G1 + jB1
G2 + jB2
vi
G1 + jB1
(a)
G2 jB2
Feedback
patch
(b)
Figure 29. (a) A transmission line model for a patch antenna, and
(b) its circuital equivalent as resonance.
1 mn
(ex km cos kn x sin km y ey kn sin kn x cos km y)
j abt
a
(1 j 0.636 ln k0 t)
0 Z0
FET
Gate
Drain
1
LC
Cslit
D
G
ZL
LT
1
1
=
Yi
2G1
ACTIVE ANTENNAS
x=0
x=d
x = 2d x = 3d
x = 4d
x = 5d
N 1
x = Nd
E t ( ) = E e
N1
e inkd sin e i n
n=0
The sum is generally referred to as the array factor. The intensity, then, in the direction is
Ri0 RT
cos 0t
m
AF
AF
kd = /2
kd = 2
kd =
2
N1
1 e iNkd sin 2
ind sin
e
=
n=0
1 e ikd sin
(6)
2 v Ri RT
v
mv2
1
+ 02 v = A cos i t
t 2
L
Ri RT t
2
N1
iknd sin i n
It ( ) = Ie
e
e
n=0
n = nkd sin
205
(a)
(b)
AF
(c)
Figure 33. Plots of the array factor of Eq. (6), where (a) N 1, (b)
N 5 and kd /2, , and 2, and (c) N 10 and kd .
206
ACTIVE ANTENNAS
to have true time delay, which means that one would like to
have as much as a phase shift between adjacent elements.
I do not think that the frequency-shifting technique can
achieve this either. Work, however, continues in this exciting area.
APPLICATIONS OF AND PROSPECTS FOR ACTIVE ANTENNAS
Perhaps the earliest application of the active antenna concept
(following that of Hertz) was aimed at solving the smallantenna problem. As we recall, an antenna can be modeled
(roughly) by a series RLC network with the R representing
the radiation resistance. The input impedance of such a combination is given by
Zi =
1 2 /02 + jRC
jC
1
LC
Antenna
R
Q1
Amplifier
ACTIVE ANTENNAS
207
208
ACTIVE ANTENNAS
Active device
(a)
G
Source bias
Drain bias
Source bias
Gate bias
Source bias
(b)
Figure 35. Schematic depiction of (a) the active surface of a grid
oscillator and (b) a breakout of an internal region of the grid showing
the active device placement relative to the bias lines.
drain or, for the gate bias CPW, the horizontal ac gate-to-gate
connection lines. Indeed, the grid is known to lock strongly
between the rows and more weakly between columns. As adjacent row elements are sharing a patch radiator, this behavior should be expected.
In a sense, this strong locking behavior of the grid is both
an advantage and a disadvantage. It is advantageous that the
grid is compact (element spacing can be /6) and further
that it is easy to get the rows to lock to each other. However,
the compactness is also a disadvantage in that it is quite hard
to get any more functionality on the grid. Much effort has
been made in this area to generate functionality by stacking
various grid-based active surfaces such as amplifying surfaces, varactor surfaces for frequency shifting and modulation, doubling surfaces, etc. A problem with stacking is, of
course, diffraction as well as alignment. Alignment tolerance
adds to complexity. Diffraction tends to ease alignment tolerance, but in an inelegant manner. A 100-transistor array with
/6 spacing will have an extent of roughly 1.5 per side. As
the diffraction angle is something like the wavelength divided
by the array diameter, the diffraction angle for such an array
is a good fraction of a radian. One can say that grids are
quasi-optical, but in optics one generally doesnt use apertures much smaller than a millimeter (center optical wavelength of micrometers), for which the diffraction angle would
be roughly a thousandth of a radian. As far as pure combining
efficiency goes, grids are probably the optimal solution. However, more functionality may well be hard to obtain with
this solution.
As we have mentioned, there are a number of techniques
for steering being investigated. There seems to be less work
on modulation, and I do not know of any simultaneous steering of modulated beams to date. Although the field of active
antennas began with the field of radio frequency, it still
seems to be in its infancy. However, as I hope this article has
brought across, there is a significant amount of work ongoing,
and the field of active antennas will grow in the future.
BIBLIOGRAPHY
1. J. A. Navarro and K. Chang, Integrated Active Antennas and Spatial Power Combining, New York: Wiley, 1995.
2. R. A. York and Z. B. Popovic (eds.), Active and Quasi-Optical
Arrays for Solid-State Power Combining, New York: Wiley, 1997.
2a. H. Hertz, Electric Waves, New York: Macmillan, 1983. (This is
a book of reprints of Hertz work in the 1890s.)
3. J. Bardeen and W. Bratain, The transistor: A semiconductor triode, Phys. Rev., 74: 435, 1948.
4. W. Shockley, The theory of pn junctions in semiconductors and
pn junction transistors, Bell Syst. Tech. J., 28: 435, 1949.
5. W. Shockley, A unipolar field-effect transistor, Proc. IEEE, 40:
13651376, 1952.
6. C. A. Mead, Schottky-barrier gate field-effect transistor, Proc.
IEEE, 54: 307308, 1966.
7. R. A. York, R. D. Martinez, and R. C. Compton, Active patch antenna element for array applications, Electron. Lett., 26: 494495,
March 1990.
8. A. R. Mickelson, Physical Optics, New York: Van Nostrand Reinholt, 1992, chap. 2.
9. B. J. Hunt, The Maxwellians, Ithaca, NY: Cornell University
Press, 1991, chap. 3.
ACTIVE FILTERS
10. H. C. Pocklington, Electrical oscillations in wires, Proc. Cambridge Phil. Soc., 324333, 1897.
11. D. M. Pozar, Microwave Engineering, Reading, MA: Addison-Wesley, 1990.
12. P. E. Gray and C. L. Searle, Electronic Principles, Physics, Models
and Circuits, New York: Wiley, 1967.
13. R. E. Collin, Foundations for Microwave Engineering, 2nd ed.,
New York: McGraw-Hill, 1992.
14. K. Chang, Microwave Solid-State Circuits and Applications, New
York: Wiley, 1994.
15. K. Y. Chen et al., Analysis of an experimental technique for determining Van der Pol parameters of a transistor oscillator, IEEE
Trans. Microw. Theory Tech., 46: 914922, 1998.
209
40. J. R. Copeland and W. J. Robertson, Antenna-verters and antennafiers, Electronics, pp. 6871, 1961.
18. D. M. Pozar and D. H. Schauber (eds.), Microstrip Antennas, Piscataway, NJ: IEEE Press, 1995.
42. W. J. Robertson, J. R. Copeland, and R. G. Verstraete, Antennafier arrays, IEEE Trans. Antennas Propag., 2: 227233, 1964.
ALAN R. MICKELSON
University of Colorado
32. R. A. York and R. C. Compton, Coupled-oscillator arrays for millimeter-wave power-combining and mode-locking, IEEE MTT-S
Int. Microw. Symp. Digest, 1992, pp. 429432.
33. P. S. Hall and P. M. Haskins, Microstrip active patch array with
beam scanning, Electron. Lett., 28: 20562057, 1992.
34. P. S. Hall et al., Phase control in injection locked microstrip ac-
ANTENNA ACCESSORIES
531
Attenuator
R2
R1
R3
Zi
Zo
ANTENNA ACCESSORIES
Accessories in antennas are devices for radiating or receiving
electromagnetic waves. Their main purpose is to create
means for the perfect operation of antennas. The accessories
interconnect these antennas to the transmission and reception systems efficiently and safely. An electromagnetic coupling generally exists between the antennas and the accessories, and one can explore them to improve antenna
performance. In this way, the accessories are as fundamental
as the antenna itself, whose behavior depends on appropriate connections.
Among several kinds of accessories, the following are most
commonly used: attenuators, baluns, chokes, coax lines, insulators, lightning arresters, planar structures, phase shifters,
twin-lead. These components are described in the next sections.
Design of an Attenuator
The designer has two concerns in designing an attenuator:
1. attenuation in dB
2. input and output impedances
The attenuation is given by Eq. (1):
ATT = 10 log10 N
(1)
where ATT is attenuation in decibels and N is the ratio between the power absorbed by the circuit (from the generator)
and the power delivered to the load.
The design starts from the desired input and output impedances and from the value of the ATT obtained in Eq. (1).
Equations (2), (3), and (4) give the resistances of the T circuit:
ATTENUATORS
Attenuators are circuits designed to introduce a known loss
between input and output ports (1). The power ratio, expressed in decibels, between the input and output represents
the loss in these circuits. The main use of attenuators is in
measuring standing wave ratio (SWR) in antennas and the
transmission coefficient. In SWR measurement (2), one adjusts the attenuation value to maintain equal outputs in the
stationary wave detector at the maximum and minimum
points. Then the SWR in dB is equal to the difference between
the readings of the attenuator.
The external circuits connected to the input and output
ports of the attenuator should present purely resistive impedances. These are always matched to the input and output impedances of the component. Therefore, the resistors always
constitute the attenuator circuit. In general, these resistive
circuits use the T and circuit topologies. Figure 1 shows
the T topology, where R1, R2, and R3 form the circuit. The
external circuitry presents purely resistive impedances (Zi,
Zo) at the input and output of the attenuator. Figure 2 shows
the configuration of the section. If the input and output
impedances are the same (Zi Zo), the circuits T and
become symmetrical, that is, R1 R2.
(2)
(3)
(4)
Attenuator
R3
Zi
R1
R2
Zo
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
532
ANTENNA ACCESSORIES
very thin, tapered, resistive card. The attenuator device introduces this card in a section of the slotted guide. The adjustment of the penetration depth in the slot allows controlling
the dissipation of power and the desired attenuation. This
type of attenuator has a complex attenuation varying with
penetration depth and frequency.
Attenuator
R1/2
R2/2
R3
Zi
Zo
BALUNS
R1/2
R2/2
Figure 3. Balanced attenuator of the H type. The R1 and R2 resistors from the T circuit are distributed in the lower arms.
Equations (5), (6), and (7) give the resistances for a type
circuit:
R3 = 1/2(N 1)(Zi Zo /N)1/2
(5)
(6)
(7)
Equations (2), (3), (4) and Eqs. (5), (6), (7) are also valid for
symmetrical circuits (it is sufficient that Zi Zo).
Attenuators T and are applied to unbalanced systems, such as coaxial cables. Bifilar lines feed many antennas
that are intrinsically balanced, such as dipoles. In this case,
the circuits T and change into their balanced versions
and they are called H and O sections, respectively. Figure
3 shows the H circuit resistor configuration, and Fig. 4 illustrates the O circuit configuration. With respect to the circuits T and , there is a distribution of the resistors R3
and R1, R2 in the lower branch. All equations shown previously are valid for this case. However, it is necessary to take
the half of the values of the respective resistors and distribute
them in the upper and lower branches of their circuits.
Attenuators in Waveguides
Waveguides usually feed high-frequency antennas. In this
case, (p. 262, Ref. 2) the attenuators are built starting from a
Attenuator
R3/2
Zi
Ri
R2
Zo
R3/2
Z1
A
;;;
Z2
Z3
(a)
;;;;
(b)
ANTENNA ACCESSORIES
533
Figure 6. (a) Dipole antenna with symmetrical feeding. (b) Monopole antenna
fed in the base. (c) Dipole antenna with
asymmetrical feeding.
(c)
load (seen at the input of the twin-lead line). The upper limit
is at the resonance point of the space of the windings. A technique used for increasing the inductive effect is to wind the
cable in ferrite cores, which present a high impedance level
on a wider frequency band. Careful designs of structures of
this type allow their operation in frequency band ratios up to
10 : 1 (5).
The baluns most used are those based on quarter-wavelength short-circuited line sections. Such an arrangement is
shown in Fig. 11. In this way, high impedance is obtained
from point B to the external side of the outer mesh of the
coaxial line. This prevents current flow into the outer surface
of the external mesh of the cable.
In practice, the characteristic impedance of the transmission line is usually real, whereas that of the antenna element
is complex. It is frequently desirable to operate the balun at
a length other than a quarter-wavelength, in order to take
advantage of the shunt reactance presented by the balun to
the load, for impedance matching purposes.
Among many coupling matching networks that can be used
to connect the transmission line to the antenna, we can introduce a quarter-wavelength transformer (see Fig. 23). If the
impedance of the antenna is real, the transformer is attached
directly to the load. However, if the antenna impedance is
complex, the transformer is placed at a distance l away from
the antenna. The distance l is chosen so that the impedance
towards the load, at l, is real.
If the transformer is a 2-wire line section, and the line is
a coaxial cable, obviously the coaxial line cannot be connected
directly to the 2-wire line. In this case, a balun may be used
to connect the transformer to the coaxial line.
RADIO-FREQUENCY CHOKES
A radio-frequency choke coil is an inductance designed to offer a high impedance to alternating currents over the frequency range for which the coil is to be used. This result is
obtained by making the inductance of the coil high and the
distributed capacitance low. The result is that the inductance
;;;;
;
;;;;
A
A
B
ZL
ZL
ZS
Zg
ZS
534
ANTENNA ACCESSORIES
ZL
;;;;
B
/4
ZS
/4
(a)
(b)
Figure 11. (a) Balun bazooka. (b) Cross section of the balun.
Breakdown occurs in an air-filled coaxial line at atmospheric pressure when the maximum electric field Em reaches
a value of approximately 2.9 106 V/m. The average power
P that is transmitted in a matched coaxial line under these
conditions is given by
COAXIAL CABLE
2
P = (Em
/480)b2[ln(b/d)/(b/d)2] W
IA
c /2(b + d)
IB
(12)
INSULATORS
A material is an insulator if this is its most dominant characteristic. This does not mean that other electric effects, such
as conductivity and magnetism, are not present, but they are
just less significant. From a macroscopic point of view, the
constitutive relationship of the material, specifically in the
case of insulators (dielectrics), is given by
E
D = E
(13)
where E is the electric field intensity vector and D is the electric flux density vector.
ZL
;;;;;
F
(11)
(9)
(10)
ZS
Figure 12. Geometry of a coaxial cable. A dielectric of relative permittivity r fills the space between the inner and outer conductors.
ANTENNA ACCESSORIES
(14)
(15)
(16)
(17)
535
materials used in antenna engineering. It also lists the breakdown field for some materials.
LIGHTNING ARRESTERS
Lightning is an atmospheric phenomenon with potentially
harmful consequences. It is caused by the accumulation of
electric charges in a cloud and the consequent discharge to
the terrestrial surface. This effect occurs mainly on structures
that offer favorable conditions to discharge. The best protection against atmospheric discharges is the enclosure by a
grounded conductive structure. Even so, such protection when
not expensive, sometimes becomes impracticable.
Antennas must be immune to atmospheric discharges and
maintain integrity as lightning current flows to the ground.
Protection against the direct and induced effects of lightning
has to be designed to avoid damage to operators, equipment,
and structures. For antennas, several techniques can be used,
such as
protective conductors on a radome (7) (without degrading
the performance of the antenna)
grounding the towers (always positioning the antennas
below the tower summit)
lightning arresters (mainly in HF operation, where the
antennas are relatively large structures)
Common types of lightning arresters are varistors, gas discharge devices, and semiconductors. No one type is suitable
for all applications, and each may be combined with another
into a hybrid device. A simple lightning arrester can be assembled by creating a gap between the structure to be protected and the ground. These gaps are adjusted to minimum
width such that no arcing occurs when the transmitter is operating.
PHASE SHIFTERS
/0 at
108 Hz
/ at
108 Hz
Breakdown
field at
25C (V/m)
Ceramics
Aluminum oxide
(alumina)
Porcelain
(dry process)
8.8
0.00030
5.04
0.0078
0.00020
0.00020
47.2 106
39.4 106
0.034
15.9 106
Plastics
Polyethylene
Polytetrafluorethylene
(Teflon)
2.26
2.1
Adhesives
Epoxy resin
(Araldite CN-501)
3.35
Fused quartz
a
3.78
0.00020
16.1 106
Shifters can provide fixed or variable phase shifts. Fixedphase shifters are usually extra transmission line sections of
certain lengths to shift the phase with respect to the reference
line. Variable shifters use mechanical or electronic means to
536
ANTENNA ACCESSORIES
2
Figure 13. PIN switching-circuit phase-shifter representation. The
application of the bias current results in a change of the electric
length of the line. The resulting shift is the difference between the
electrical angle in the direct and reverse polarized cases.
switching-circuit shifters are common to several manufacturing technologies, such as microstrip, coplanar guides and
lines, slotline, waveguides, and finlines. The diodes act as binary switches along the line (Fig. 13). With the bias applied,
the diode behaves as a short circuit for the incoming wave.
Therefore, the phase-transmitted signal has two possible
states, determined by which diode is polarized. A variable
phase shift results from mounting several diodes on the line.
The returning wave enters a hybrid and then the antenna.
PLANAR TECHNOLOGY
Planar transmission lines are the basic transmission media
for microwave integrated circuits when hybrid or monolithic
technology is used. Figure 14 shows the cross sections of three
basic planar transmission lines: stripline, microstrip, and
slotline. Several other configurations are possible, such as
listed in Ref. (10). All configurations are variations of the
three main types. Each offers certain advantages, depending
on the circuit type and its application. Some of these configurations are the suspended stripline, the suspended microstrip,
the inverted microstrip, the coplanar guide, and the coplanar
stripline. Planar designs using special substrates have been
applied at frequencies up to 100 GHz.
A designer can produce several microwaves circuits and
electric functions with planar transmission lines. Therefore,
planar technology constitutes a very important technology for
developing antenna accessories in the microwave range.
These components, called distributed circuits, are generally of
the order of a wavelength or more (in the propagation medium). Therefore, in the lower microwave range, these circuits
can have large dimensions. In this case, one uses lumped components.
Several lumped components can be obtained starting from
planar technology. These components are called microwave
integrated circuits (MIC). Figure 15 illustrates lumped and
distributed circuits in planar technology.
Following are the most important design parameters for
planar transmission lines:
effective dielectric constant
characteristic impedance
dispersion
;;
;;
;;
ANTENNA ACCESSORIES
537
Strip conductor
Ground
plane
Dielectric
substrate
Stripline-TEM mode
Ground
plane
(a)
(a)
Strip conductor
Dielectric
substrate
Ground
plane
Ground plane
Ground plane
Slot
Dielectric
substrate
w
b
Slotline- non-TEM mode
(c)
Figure 14. Basic planar transmission lines. (a) StriplingTEM
mode. (b) Microstrip linequasi-TEM mode. (c) Slotlinenon-TEM
mode.
losses
working frequency limitations
Several antenna accessories are manufactured for planar
technology. Examples are directional couplers, hybrid rings,
power dividers, and impedance matchers (11,12). These components are used as matching, feeding, and sampling networks in antenna systems.
The effect of the electromagnetic coupling between two
parallel transmission lines (sufficiently near each other) constitutes the design framework of several microwave devices
(especially for antennas). An instance is the directional coupler, which has wide application in circuits and telecommunication systems. It performs several electric functions, such as
separating and combining signal power (received or delivered
from/to antennas). The directional coupler also allows sampling power and separating incident and reflected waves of a
system. This separation is used for power control or performance measurements.
Couplers are either conductively or electromagnetically
coupled transmission line circuits with three, four, or more
ports. In general, a directional coupler is a four-port device.
Two of its ports are mutually decoupled with respect to the
Coupled
output 3
port
Input
port
a
d
Isolated
port
Direct
2 output
port
538
ANTENNA ACCESSORIES
Isolated
4
port
Coupled
port
Direct
port
Y3
Y1
Input
port
3
Y2
/4
Y4
1
/4
(a)
Figure 19. Hybrid ring (courtesy of the Department of Electrical Engineering of the University of Braslia).
Bridging
wire
connection
w2
Isolated
Input
s
w1
(b)
Figure 18. (a) Branch line coupler structure. (b) Branch line coupler
in a phase-shifter circuit (courtesy of the Department of Electrical
Engineering of the University of Braslia).
Coupled
/8
/8
Direct
ANTENNA ACCESSORIES
From
source
Zl
Zt
To
load
Zl
539
Patch
Feed
Z sZ l
Zt =
/4
(a)
(b)
/4
Patch
Figure 21. Lange coupler in tandem topology (courtesy of the Department of Electrical Engineering of the University of Braslia).
signal. The output signals in the direct and coupled ports are
180 out of phase.
Another topology of the directional coupler is the Lange
coupler (Fig. 20) which is well adapted to microstrip technology (13). It is an interdigital structure with superior performance compared with the parallel line coupler. Its bandwidth
can be greater than one octave. The Lange coupler associated
with tandem topology (Fig. 21) allows obtaining power coupling near 3 dB with a combination of two lower order couplers, whose design dimensions are feasible.
Power dividers are used in feeds and performance measurements of antennas. Directional couplers and hybrid rings
can be used as power dividers. The branch line coupler output
signals are in phase quadrature (90 out of phase). Output
signals of hybrid rings are in opposite phases. However, in
several microwave applications, such as parallel feeds for
phased array antennas, the input has to be divided into an
arbitrary number of signals with equal power and in phase. A
symmetrical power divider of n branches provides successive
division of the input signal into n signals with the same
power and phase at all frequencies. This power divider can
also be used as a signal combiner by inverting the input and
output ports.
There are several kinds of power dividers/combiners. One
of them is the two-branch power divider, shown in Fig. 22. It
Output 1
2 Zc
Zc
Zc
Input
2 Zc
Zc
Output 2
(a)
Output 1
2 Zc
Zc
Zc
R = 2 Zc
Input
2 Zc
Zc
Output 2
(b)
Figure 22. Unmatched (a) and matched (b) two-way power dividers.
Zs
Zl
k3
Zl
(c)
l3
Z5
Z3
l2
Z2
l1
Z1
Stripline
Strip
Microstrip
Substrate
(d)
540
ANTENNA ACCESSORIES
2r
TWIN LEAD
The bifilar line (twin-lead) is a two-wire parallel conductor
line, which carries power from the generator to the antenna.
Each wire carries equal and opposite currents (180 out of
phase). Because the wires are spaced by a certain distance,
however, their radiated fields do not cancel out completely. If
the two wires were in the same place in space, there would
be no radiated field. However, because this is not possible,
there will be a certain amount of radiation loss. Keeping the
distance between the parallel wires small (typically on the
order of 1% of the wavelength of the radio wave) reduces this
loss. The spacing between conductors is a constraint of the
physical limitations of line construction. Parallel conductor
lines, open-wire lines, two-wire lines, two-wire cable and twowire ribbon cables are also bifilar lines.
Types of Bifilar Lines
The bifilar lines are usually supported a fixed distance apart.
Insulating rods, called spacers, or molded plastics, such as
polystyrene, provide the support. The spacers have to be
placed at short intervals to prevent the wires from moving
apart from each other. The lines are air-insulated, because
this is the medium between them. The spacers have little effect on the impedance behavior of the line. Flexible dielectrically separated (molded) lines have several advantages
over air-insulated types. In this type, a plastic coating similar
to a ribbon surrounds the conductors. These lines maintain
uniform spacing between conductors. They are also less
bulky, weigh less, and are easier to install. One common example of a flexible dielectric bifilar lines is the television receiving cable (called ribbon cable or two-wire cable).
d
Figure 24. The geometry of bifilar lines. In the molded plastic casing, the dielectric surrounding the lines has the shape of a ribbon.
Because most of the field is inside the ribbon, the line is less susceptible to external influences.
1
Z0 = cosh
d
2r
120
1
= cosh
r
d
2r
(18)
A(dB/m) = 8.686
1
r
d/2r
(d/2r)2 1
+
(19)
= /v
(20)
where v is the phase velocity in the medium surrounding the conductors. In a ribbon transmission line,
r is 2.56 with (/ ) 0.7. If the wire is made of copper,
then is 58 MS/m. In this case, because most of the
field is inside the dielectric, the relative dielectric constant r is approximately 2.56.
BIBLIOGRAPHY
1. Reference Data for Radio Engineers, 7th ed., Indianapolis, IN:
H.W. Sams, 1985.
2. R. E. Collin, Foundations for Microwave Engineering, 2nd ed.,
New York: McGraw-Hill, 1992.
3. A. W. Rudge et al., The Handbook of Antenna Design, London:
Peter Peregrinus, 1986, Vols. 1 and 2.
4. H. Jasik, Antenna Engineering Handbook, New York: McGrawHill, 1961.
ANTENNA ARRAYS
5. W. L. Weeks, Antenna Engineering, New York: McGraw-Hill,
1968.
6. S. Ramo, J. R. Whinnery, and T. Van Duzer, Fields and Waves
in Communication Electronics, 3rd ed., New York: Wiley, 1994,
pp. 678682.
7. R. H. Golde, Lightning, New York: Academic Press, 1981, Vols. 1
and 2.
8. D. M. Pozar, Microwave Enginering, Reading, MA: Addison-Wesley, 1990.
9. J. Uher, J. Bornemann, and U. Rosenberg, Waveguide Components for Antenna Feed Systems. Theory and CAD, Norwood, MA:
Artech House, 1993.
10. B. Bhat and S. Koul, Stripline-Like Transmission Lines for Microwave Integrated Circuits, New York: Wiley, 1989.
11. E. A. Wolff and R. Kaul, Microwave Engineering and Systems
Applications, New York: Wiley, 1988.
12. L. A. Trinogga, G. Kaizhou, and I. C. Hunter, Practical Microstrip
Circuit Design, London: Ellis Horwood, 1991.
13. J. Lange, Interdigited stripline quadrature hybrid, IEEE Tran.
Microw. Theory Tech., MTT-17: 11501151, 1969.
14. G. Hall (ed.), The ARRL Antenna Book, 15th ed., Newington, NJ:
The American Radio League, 1988, pp. 24.1524.25.
PLINIO R. G. ALVES
LEONARDO R. A. X. MENEZES
PAULO H. P. CARVALHO
FRANKLIN C. SILVA
University of Brasilia
541
ANTENNA ARRAYS
Consisting of two to thousands of antenna elements, an array antenna presents the ultimate in exible antenna pattern control. This capability includes electronic scanning,
planar or conformable apertures, the ability to transmit or
receive multiple shaped patterns, and adaptive control for
jammers, clutter, and multipath suppression and for the
reduction of cosite interference. Arrays benet from technological advances, and they are the key technology drivers
for solid state T/R (transmit/receive) modules, advanced
signal processing, and photonic technology.
Antenna arrays are often simply called phased arrays,
a term that refers to the progressive phase shift introduced to scan the beam. We will use this term throughout this section, although the beam can be scanned by either time delay devices or phase shifters, depending on the
required systems bandwidth. The distinction will be addressed later.
The basic principle behind the operation of the phased
array, as illustrated in Fig. 1, is that the RF power be divided among a number of elements by a power divider, with
each element signal shifted appropriately in phase or in
time. Relative time delay is portrayed in the sketch by circular phase fronts emanating from each element of the array, with signals either radiated at the same time [Fig. 1(a)]
or delayed in time by an increasing amount from left to
right so that the rightmost signal radiates last. The gure
shows how radiation from each element adds in space so as
to create an outgoing wave with an appropriate scan angle.
Although not shown in the gure, the array power divider
usually provides equal line lengths from the source to each
element because this leads to optimum system bandwidth.
The power divider can also be used to control the signal
amplitude at each element. This unequal power division
is called amplitude tapering, and it provides for sidelobe
control. At broadside, with element spacing chosen properly, the array directivity is that of the broadside aperture,
where A is the aperture area and a is the aperture efciency, which depends on the arrays taper design. The
scanned beam of Fig. 1(b) will have reduced directivity and
(usually) additional losses that further reduce the array
gain.
ARRAY ARCHITECTURE AND COMPONENTS
Figure 1 also indicates the basic components that make up
a phased array. They consist of a power divider/combiner,
phase shifters, and radiating elements. This simple sketch
understates the tremendous number and variety of components that make up array technology, and it also understates the importance of an overriding array architecture
aimed at satisfying the system needs. The following paragraphs will illustrate some of these issues.
Architecture is the character and style of building of a
structure or system. The term array architecture implies
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Antenna Arrays
sive and often lighter. Figure 3(a) shows that with one such
space feed, the array is used as an active lens. Array elements at the back face of the array receive the RF power
from the horn feed. Phase shifters installed in the path
between array back and front faces correct for the nearly
spherical wavefront of the incident wave and then add the
progressive phase distribution to scan the beam. Figure
3(b) shows the reectarray concept in which the array is
space-fed from the front. Again the elements at the front
face of the array receive the distributed power, and phase
shifters provide the necessary phase correction and scanning control. The back face of the array is short-circuited,
so the RF signal goes through the phase shifter twice before it radiates. This tends to make reectarrays somewhat
more lossy and the phase shifter more tolerance sensitive
than lens arrays. This feature, and undesirable feed blockage as indicated in Fig. 3(b), for low-angle radiation have
made space-fed lens arrays the more popular choice for
most space-fed systems.
Phase Shifters and Time Delay Units
Phase shifters are devices that change the phase of a signal traveling down the feed transmission line. Usually the
devices have discrete phase states, so one measure of the
quality of the phase shifter is the number of bits, or digital
states, by which it approximates a continuous phase dis-
Antenna Arrays
Figure 2. Array and subarray congurations. (a) Column subarrays with brick construction. (b) Area subarrays with tile construction.
Antenna Arrays
Figure 5. Array elements. (a) Microstrip patch element. (b) Proximity coupled microstrip element. (c) Stripline fed cavity backed
slot element. (d) Strip line fed dipole element. (e) Flared notch
element fed by stripline center conductor.
ARRAY THEORY
Array Antenna Patterns
The far-eld pattern of an array of N elements in the generalized geometry of Fig. 6 can be written in terms of the
incident transmission line voltage signals an and the element locations r n as
for
Antenna Arrays
for
This beam angle moves, or squints, as a function of frequency as indicated in Fig. 7, with the beam peak farthest from broadside at the lowest frequency and nearest
to broadside for the highest frequency.
This squint angle change can be interpreted in terms of
a fractional bandwidth by assuming an array beamwidth
of u and assuming that the beam is placed exactly at the
angle 0 at center frequency f0 . Then dening the upper
and lower usable frequencies to be those at which the gain
Antenna Arrays
Figure 8. Array factors illustrating grating lobe phenomenon. Solid curve: column array of 64 half
wavelength spaced elements scanned to 45 degrees. Dashed curve: column array of 16 elements
with 2 wavelength spacing.
Antenna Arrays
subject to
Figure 9. Array mutual coupling. (a) Coupling between array elements. (b) Element pattern P() and reection coefcient magnitude R of center element of unloaded waveguide array (B/ = a/
= 0.4) after Wu (3).
Antenna Arrays
thus:
where
The beam width along the scan plane thus broadens like
sec 0 as the array is scanned in .
Accompanying this beam width increase is a decrease
in array directivity so that the directivity is given in terms
of the broadside directivity DB as
ARRAY SYNTHESIS
Many useful pattern synthesis techniques for planar or linear antenna arrays follow directly from existing methods
developed for aperture or continuous one-dimensional antennas. This is so for several reasons. First, as long as the
elements are closely spaced and grating lobes well out of
the radiating region, the array periodicity does not signicantly alter the pattern structure. Second, the distinctions
that do exist come from the mutual coupling and are evident in array edge effects, or equivalently from the observed different element patterns across the array. As long
as the elements support only the single-mode elds, these
issues do not alter the synthesis procedure, since one can
synthesize in terms of the currents and aperture elds that
create radiation, or in terms of the measured or computed
array element patterns, and then include mutual coupling
to evaluate the necessary applied excitation.
The basis for most aperture synthesis is the Fourier
transform relationship between aperture eld and far eld
for a continuous aperture. If the arrays are large and the
elements closely spaced, this procedure is not sensitive to
the discretization or edge effects, and the method is quite
accurate. The transform method is also especially convenient because of its application to arrays that are not periodic and for arrays conformal to gently curved geometries.
Arrays periodic in one or two dimensions have far-eld
patterns describable by discrete Fourier transform pairs.
In one dimension the array factor at wavelength is written
where the sum is taken symmetrically about the array center. The coefcients an are the array element excitation and
are given from orthogonality as
Antenna Arrays
Figure 10. Low sidelobe sum and difference pattern synthesis. (a) Taylor sum pattern with 30 dB (n = 6) pattern.
(b) Bayliss difference pattern with 30 dB (n = 6) pattern.
as improvements to the equal ripple method of DolphChebyshev, and result in more realizable aperture distributions, improved gain and other advantages. Figure 10(ab)
shows the array factors for 32 element arrays with 30 dB
Taylor and Bayliss distributions. Note that the rst sidelobe in both cases is very close to 30 dB with respect to
the pattern maximum. In general, the discretizing of continuous distributions introduces errors in the synthesized
pattern, and these are more signicant for small arrays or
for arrays that are forced to have very low sidelobes. Space
precludes giving a detailed description of these procedures,
but they are described in detail in a number of references.
Usually discretizing the continuous distribution is not a
problem, but when it is, there are a number of iterative
techniques to converge to the original desired pattern. Notable among these is the work of Elliott (10).
10
Antenna Arrays
Figure 11. Circular array with elements removed). (a) Geometry (dashes show elements removed. (b) Desired Taylor pattern of
lled array. (c) Pattern of thinned array.
Antenna Arrays
11
been built in laboratory models, and will play a role in providing ultrawideband multifunctional arrays.
Conformality and exible control are the two features
unique to array antennas, and the number of new applications continues to increase to ll demands in the two areas. Conformal scanning arrays provide high-gain antenna
coverage from complex platforms, and when combined with
elements of the increased exibility, arrays will become especially good candidates for commercial applications such
as in wireless communications and automotive radar. As
the cost of solid-state modules continues to decrease, more
applications will be lled by this high-end antenna technology.
Time Delay for Wide Band Arrays:
12
Antenna Arrays
Figure 15. Ideal subarray pattern for time delay steered arrays.
Antenna Arrays
13
14
Antenna Arrays
BIBLIOGRAPHY
1. W. E. Hord Microwave and millimeter wave ferrite phase
shifters, Microwave Journal State of the Art Reference, 32:
(September) 1989.
2. A. K. Sharma Solid state control devices: State of the art, Microwave Journal State of the Art Reference, 32: (September)
1989.
3. C. P. Wu Analysis of nite parallel plate waveguide arrays,
IEEE Trans. Antennas Propag., AP-18 (3): 328334, 1970.
4. P. M. Woodward A method of calculating the eld over a plane
aperture required to produce a given polar diagram, Proc. IEE
(London), Part 3A, 93: 15541555, 1947.
5. J. Butler R. Loe Beamforming matrix simplies design of electronically scanned antennas, Electronic Design, 9: 170173,
(12 April) 1961.
6. S. A. Schelkunov A mathematical theory of linear arrays, Bell
System Tech J., 80107, 1943.
7. C. L. Dolph A current distribution for broadside arrays which
optimizes the relationship between beamwidth and sidelobe
level, Proc. IRE, 34(June): 335345, 1946.
8. T. T. Taylor Design of line source antennas for narrow
beamwidth and low sidelobes, IEEE Trans. Antennas Propag.,
AP-3: 1628, (January) 1955.
9. E. T. Bayliss Design of monopulse antenna difference patterns
with low sidelobes, Bell System Tech. J., 47: 623640.
10. R. S. Elliott On discretizing continuous aperture distributions,
IEEE Trans. Antennas Propag., AP-25: 617621, (September)
1977.
11. H. R. Orchard R. S. Elliott G. J. Stern Optimizing the synthesis
of shaped antenna patterns, IEE Proc. (London) Part H (1):
6366, 1984.
12. F. Ares et al. Application of genetic algorithms and simulated
annealing technique in optimising the aperture distributions
of antenna array patterns, Electronic Lett., 32 (3): 148149,
1996.
13. M. K. Skolnik J. W. Sherman III F. C. Ogg, Jr. Statistically Designed Density-Tapered Arrays, IEEE Trans. Antennas Propag., AP-12: 408417, (July) 1964.
14. R. J. Mailloux, S. G. Santorelli and T. M. Roberts, Wideband
Arrays using Irrregular (Polyomino) shaped Subarrays, electronics Letters, Vol.42, No.18,Aug. 2006, pp. 10191020
15. R. J. Mailloux, An Overlapped Subarray for Limited Scan Applications, IEEE Trans. AP-22, No.3,May 1974, pp. 487489
ROBERT J. MAILLOUX
University of Massachusetts,
Dept. of Electrical and
Computer Engineering,
Amherst, MA
, where
The first section of this article introduces several antenna patterns, giving the necessary definitions
and presenting the common types. The field regions of an antenna are also pointed out. The most common
reference antennas are the ideal isotropic radiator and the very short dipole. Their fields are used to show the
calculation and meaning of the different parameters of antennas covered in this article. The second section
begins with a treatment of the Poynting vector and radiation power density, starting from the general case of an
electromagnetic wave and extending the definitions to a radiating antenna. After this, radiation performance
measures such as the beam solid angle, directivity, and gain of an antenna are defined. In the third section
the concepts of wave and antenna polarization are discussed. Finally, in the fourth section, a general case of
antenna pattern calculation is considered, and numerical solutions are suggested for radiation patterns that
are not available in simple closed-form expressions.
Fig. 1. Spherical coordinate system for antenna analysis. A very short dipole is shown with the directions of its nonzero
field components.
=
distance from center of dipole to observation point, (m) and 0 = permittivity of free space = 8.85
pF/m.
Note that E and H are in time phase in the far field. Thus, the electric and magnetic fields in the far
field of the spherical wave from the dipole are related in the same manner as in a plane traveling wave. Both
Fig. 2. Radiation field pattern of far field from an ideal (very short) dipole: (a) three-dimensional pattern plot, (b) E-plane
radiation-pattern polar plot, and (c) H-plane radiation-pattern polar plot (HPBW, H-plane beamwidth).
are also proportional to sin . That is, both are at a maximum when = 90 and a minimum when = 0 (in the
direction of the dipole axis). This variation of E or H with angle can be presented by a field pattern, shown
in Fig. 2, the length r of the radius vector being proportional to the value of the far field (E or H ) in that
direction from the dipole. The pattern in Fig. 2(a) is the three-dimensional far-field pattern for the ideal dipole,
while the patterns in Fig. 2(b, c) are two-dimensional and represent cross sections of the three-dimensional
pattern, showing the dependence of the fields on the angles and .
All far-field components of a very short dipole are functions of I0 , the dipole current; L/, the dipole length
in wavelengths; 1/r, the distance factor; jej(tr) , the phase factor; and sin , the pattern factor, which gives
the variation of the field with angle. In general, the expression for the field of any antenna will involve these
factors.
For longer antennas with complicated current distribution the field components generally are functions of
the above factors, which are grouped into the element factor and the space factor. The element factor includes
everything except the current distribution along the source, which is the space factor of the antenna. For
example, consider the case of a finite dipole antenna. The field expressions are produced by dividing the
antenna into a number of very short dipoles and summing all their contributions. The element factor is equal
to the field of the very short dipole located at a reference point, while the space factor is a function of the
current distribution along the source, the latter usually described by an integral. The total field of the antenna
is given by the product of the element and space factors. This procedure is known as pattern multiplication.
A similar procedure is also employed in array antennas, which are used when directive characteristics
are needed. The increased electrical size of an array antenna due to the use of more than one radiating element
gives better directivity and special radiation patterns. The total field of an array is determined by the product
of the field of a single element and the array factor of the array antenna. If we use isotropic radiating elements,
the pattern of the array is simply the pattern of the array factor. The array factor is a function of the geometry
of the array and the excitation phase. Thus, changing the number of elements, their geometrical arrangement,
their relative magnitudes, their relative phases, and their spacing, we obtain different patterns. Figure 3 shows
some of characteristic patterns of an array antenna with two isotropic point sources as radiating elements,
using different values of the above quantities, which produce different array factors.
Common Types of Radiation Patterns. An isotropic source or radiator is an ideal antenna that
radiates uniformly in all directions in space. Although no practical source has this property, the concept of the
isotropic radiator is very useful, and it is often used as a reference for expressing the directive properties of
actual antennas. It is worth recalling that the power flux density S at a distance r from an isotropic radiator
is Pt /4r2 , Pt being the transmitted power, since all the transmitted power is evenly distributed on the surface
of a spherical wavefront with radius r. The electric field intensity is calculated as
(using the relation
from electric circuits, power = E2 /, where is the characteristic impedance of free space, 377
).
On the contrary, a directional antenna is one that radiates or receives electromagnetic waves more
effectively in some directions than in others. An example of an antenna with a directional radiation pattern
is that of an ideal or very short dipole, shown in Fig. 2. It is seen that this pattern, which resembles a
doughnut with no hole, is nondirectional in the azimuth plane, which is the xy plane characterized by the set
of relations [f(), = /2], and directional in the elevation plane, which is any orthogonal plane containing
the z axis characterized by [g(), = constant]. This type of directional pattern is called an omnidirectional
pattern and is defined as one having an essentially nondirectional pattern in a given plane, which for this
case is the azimuth plane, and a directional pattern in any orthogonal plane, in this case the elevation plane.
The omnidirectional patternknown also as broadcast-typeis used for many broadcast or communication
services where all directions are to be covered equally well. The horizontal-plane pattern is generally circular,
while the vertical-plane pattern may have some directivity in order to increase the gain.
Other forms of directional patterns are pencil-beam, fan-beam, and shaped-beam patterns. The pencilbeam pattern is a highly directional pattern, which is used when it is desired to obtain maximum gain and
when the radiation pattern is to be concentrated in as narrow an angular sector as possible. The beamwidths
in the two principal planes are essentially equal. The fan-beam pattern is similar to the pencil-beam pattern
except that the beam cross section is elliptical in shape rather than circular. The beamwidth in one plane
may be considerably broader than in the other plane. As with the pencil-beam pattern, the fan-beam pattern
generally implies a rather substantial amount of gain. The shaped-beam pattern is used when the pattern in
one of the principal planes is desired to have a specified type of coverage. A typical example is the cosecant
pattern, which is used to provide a constant radar return over a range of angles in the vertical plane. The
Fig. 3. Three-dimensional graphs of power radiation patterns for an array of two isotropic elements of the same amplitude
and (a) opposite phase, spaced 0.5 apart, (b) phase quadrature, spaced 0.5 apart, (c) opposite phase, spaced 0.25 apart,
and (d) opposite phase, spaced 1.5 apart.
pattern in the other principal plane is usually a pencil-beam type, but may sometimes be circular, as in certain
types of beacon antennas.
In addition to the above pattern types, there are a number of special shapes used for direction finding
and other purposes. These include the well-known figure-of-eight pattern, the cardioid pattern, split-beam
patterns, and multilobed patterns whose lobes are of substantially equal amplitude. For such patterns, it is
generally necessary to specify the pattern by an actual plot of its shape or by a mathematical relationship.
Antennas are often referred to by the type of pattern they produce. Two terms, which usually characterize
array antennas, are broadside and endfire. A broadside antenna is one for which the main beam maximum is
in a direction normal to the plane containing the antenna. An endfire antenna is one for which the main beam
Fig. 4. Polar plots of a linear uniform-amplitude array of five isotropic sources with 0.5-wavelength spacing between the
sources: (a) broadside radiation pattern (0 phase shift between successive elements), and (b) endfire radiation pattern
(180 phase shift).
is in the plane containing the antenna. For example, the short dipole antenna is a broadside antenna. Figure 4
shows two cases of broadside and endfire radiation patterns, which are produced from a linear uniform array
of isotropic sources of 0.5-wavelength spacing between adjacent elements. The type of radiation pattern is
controlled by the choice of phase shift between the elements. Zero phase shift produces a broadside pattern,
and 180 phase shift (for this case where the spacing between adjacent element is 0.5 ) leads to an endfire
pattern, while intermediate values produce radiation patterns with the main lobes between these two cases.
Characteristics of simple patterns. For a linearly polarized antenna, such as a very short dipole
antenna, performance is often described in terms of two patterns [Fig. 2(b, c)]. Any plane containing the z axis
has the same radiation pattern, since there is no variation in the fields with angle [Fig. 2(b)]. A pattern taken
in one of these planes is called an E-plane pattern, because it is parallel to the electric field vector E and passes
through the antenna in the direction of the beam maximum. A pattern taken in a plane orthogonal to an E
plane and cutting through the short dipole antenna (the xy plane in this case) is called an H-plane pattern,
because it contains the magnetic field H and also passes through the antenna in the direction of the beam
maximum [Fig. 2(c)]. The E- and H-plane patterns, in general, are referred to as the principal-plane patterns.
The pattern plots in Fig. 2(b, c) are called polar patterns or polar diagrams. For most types of antennas it
is a usual practice to orient them so that at least one of the principal-plane patterns coincides with one of
the geometrical principal planes. An illustration is shown in Fig. 5, where the principal planes of a microstrip
antenna are plotted. The xy plane (azimuthal plane, = /2) is the principal E plane, and the xz plane (elevation
plane, = 0) is the principal H plane.
A typical antenna power pattern is shown in Fig. 6. In Fig. 6(a), a polar plot on a linear scale is depicted,
and in Fig. 6(b), the same pattern is shown in rectangular coordinates in decibels. As can be seen, the radiation
pattern of the antenna consists of various parts, which are known as lobes. The main lobe (or main beam or
major lobe) is defined as the lobe containing the direction of maximum radiation. In Fig. 6(a) the main lobe is
pointing in the = 0 direction. In some antennas there may exist more than one major lobe. A minor lobe is
any lobe except the main lobe. Minor lobes comprise sidelobes and back lobes. The term sidelobe is sometimes
reserved for those minor lobes near the main lobe, but is most often taken to be synonymous with minor lobe.
A back lobe is a radiation lobe in, approximately, the opposite direction to the main lobe. Minor lobes usually
Fig. 5. The principal-plane patterns of a microstrip antenna: the xy plane or E plane (azimuth plane, = /2), and the xz
plane or H plane (elevation plane, = 0).
represent radiation in undesired directions, and they should be minimized. Sidelobes are normally the largest
of the minor lobes. The level of side or minor lobes is usually expressed as a ratio of the power density in the
lobe in question to that of the main lobe. This ratio is often termed the sidelobe ratio or sidelobe level and
desired values of it depend on the antenna application.
For antennas with simple patterns, the half-power beamwidth and the sidelobe level in the two principal
planes specify the important characteristics of the patterns. The half-power beamwidth (HPBW) is defined in
a plane containing the major maximum beam, as the angular width within which the radiation intensity is
at least one-half the maximum value for the beam. The beamwidth between first nulls (BWFN) and the beam
widths 10 dB or 20 dB below the pattern maximum are also sometimes used. All of them are shown in Fig. 6.
However the term beamwidth by itself is usually reserved to describe the half-power (3 dB) beamwidth.
The beamwidth of the antenna is a very important figure of merit in the overall design of an antenna
application. As the beamwidth of the radiation pattern increases, the sidelobe level decreases, and vice versa.
So there is a tradeoff between the sidelobe ratio and beamwidth.
In addition, the beamwidth of the antenna is used to describe the resolution of the antenna: its ability to
distinguish between two adjacent radiating sources or radar targets. The most common measure of resolution
is half the first null beamwidth, which is usually used to approximate the half-power beamwidth. This means
that two sources separated by angular distances equal to or greater than the HPBW of an antenna, with a
uniform distribution, can be resolved. If the separation is smaller, then the antenna will tend to smooth the
two signals into one.
Field Regions of an Antenna. For convenience, the space surrounding a transmitting antenna is
divided into several regions, although, obviously, the boundaries of the regions cannot be sharply defined. The
names given to the various regions denote some pertinent prominent property of each region.
In free space there are mainly two regions surrounding a transmitting antenna, the near-field region and
the far-field region. The near-field region can be subdivided into two regions, the reactive near field and the
radiating near field.
The first and innermost region, which is immediately adjacent to the antenna, is called the reactive or
induction near-field region. Of all the regions, it is the smallest. It derives its name from the reactive field, which
lies close to every current-carrying conductor. In this region the reactive field, which decreases with either the
Fig. 6. Antenna power patterns: (a) a typical polar plot on a linear scale, and (b) a plot in rectangular coordinates on a
decibel (logarithmic) scale. The associated lobes and beamwidths are also shown.
square or the cube of distance, dominates over all radiated fields, the components of which decrease with the
from
first power of distance. For most antennas, this region is taken to extend over distances r < 0.62
the antenna as long as D , where D is the largest dimension of the antenna and is the wavelength (2).
For the case of an ideal or very short dipole, for which D = z , this distance is approximately one-sixth
10
of a wavelength (/2). At this distance from the very short dipole the reactive and radiation field components
are respectively equal in magnitude.
Between the reactive near-field region and far-field regions lies the radiating near-field region, which is
the region where the radiation fields dominate, but the angular field distribution still depends on the distance
from the antenna. For a transmitting antenna focused at infinity, which means that the rays at a large distance
from the antenna are parallel, the radiating near-field region is sometimes referred to as the Fresnel region, a
term taken from the field of optics. This is taken to be that between the end of reactive near-field region (r >
) and the starting distance of the far-field region (r < 2D2 /)(2).
The outer boundary of the near-field region lies where the reactive field intensity becomes negligible with
respect to the radiation field intensity. This occurs at distances of either a few wavelengths or a few times the
major dimension of the antenna, whichever is the larger. The far-field or radiation region begins at the outer
boundary of the near-field region and extends outward indefinitely into free space. In this region the angular
distribution of the field is essentially independent of the distance from the antenna. For example, for the case
of a very short dipole, the sin pattern dependence is valid anywhere in this region. The far-field region is
commonly taken to be at distances r > 2D2 / from the antenna, and for an antenna focused at infinity it is
sometimes referred to as the Fraunhofer region.
All three regions surrounding an antenna and their boundaries are illustrated in Fig. 7.
0.62
11
where E (V/m) and H (A/m) are the field vectors. Since the Poynting vector represents a surface power density
(W/m2 ), the integral of its normal component over a closed surface always gives the total power through the
surface. That is,
where P is the total power (W) flowing out of closed surface A, and dA = dA, being the unit vector normal
to surface. The Poynting vector S and the power P in the above relations are instantaneous values.
Normally, it is the time-averaged Poynting vector Sav , which represents the average power density, that
is of practical interest. It is given by
where Re stands for the real part of the complex number and the asterisk denotes the complex conjugate.
in Eq. (3) are the electric and magnetic fields written as complex numbers to include the
Note that and
change with time. That is, for a plane wave traveling in the positive z direction with electric and magnetic
field components in the x and y directions, respectively, the electric field is E = Ex0 ejT while in Eq. (1) it is E
= Ex0 . The factor 12 appears because the fields represent peak values; it should be omitted for rms values.
The average power Pav flowing outward through a closed surface can now be obtained by integrating Eq.
(3):
Consider the case that the electromagnetic wave is radiated by an antenna. If the closed surface is taken
around the antenna within the far-field region, then this integration results in the average power radiated by
the antenna. This is called radiation power, Prad , while Eq. (3) represents the radiation power density, Sav , of
the antenna. The imaginary part of Eq. (3) represents the reactive power density stored in the near field of an
antenna. Since the electromagnetic fields of an antenna in its far-field region are predominately real, Eq. (3) is
enough for our purposes.
The average power density radiated by the antenna as a function of direction, taken on a large sphere of
constant radius in the far-field region, results in the power pattern of the antenna.
As an example, for an isotropic radiator, the total radiation power is given by
12
Here, because of symmetry, the Poynting vector Si = Si (r) is taken independent of the spherical coordinate
angles and , having only a radial component.
From Eq. (5) the power density can be found:
The above result can also be reached if we assume that the radiated power expands radially in all
directions with the same velocity and is evenly distributed on the surface of a spherical wavefront of radius r.
As we will see later, an electromagnetic wave may have an electric field consisting of two orthogonal linear
components of different amplitudes, Ex0 and Ey0 , respectively, and a phase angle between them, . Thus, the
total electric field vector, called an elliptically polarized vector, becomes
which at z = 0 becomes
and
y.
The total
where is the phase lag of y with respect to x . From Eq. (9) the complex conjugate magnetic field can be
found changing only the signs of exponents.
Now the average Poynting vector can be calculated using the above fields:
It should be noted that Sav is independent of , the phase angle between the electric field components.
In a lossless medium = 0, because the electric and magnetic fields are in time phase and Ex0 /H x0 = Ey0 /H y0
= , where is the intrinsic impedance of the medium, which is real. If E =
and H =
13
The above expressions are the most general form of radiation power density of an elliptically polarized wave
or of an elliptically polarized antenna, respectively, and hold for all cases, including the linear and circular
polarization cases, that we will introduce later on.
Radiation Intensity. Radiation intensity is a far-field parameter, in terms of which any antenna radiation power pattern can be determined. Thus, the antenna power pattern, as a function of angle, can be
expressed in terms of its radiation intensity as (2,3)
where
U(, ) = radiation intensity (W/unit solid angle)
Sav = radiation density, or radial component of Poynting vector (W/m2 )
E(r,,) = total transverse electric field (V/m)
H(r,,) = total transverse magnetic field (A/m)
r = distance from antenna to point of measurement (m)
= intrinsic impedance of medium (
/square)
In Eq. (12) the electric and magnetic fields are expressed in spherical coordinates.
What makes radiation intensity important is that it is independent of distance. This is because in the far
field the Poynting vector is entirely radial, which means the fields are entirely transverse and E and H vary
as 1/r.
Since the radiation intensity is a function of angle, it can also be defined as the power radiated from an
antenna per unit solid angle. The unit of solid angle is the steradian, defined as the solid angle with its vertex
at the center of a sphere of radius r that is subtended by an area on the sphere equal to r2 . But the area of a
sphere of radius r is given by A = 4r2 , so in the whole sphere there are 4r2 /r2 = 4 sr. For a sphere of radius
r, an infinitesimal area dA on the surface of it can be written as
14
Thus, the total power can be obtained by integrating the radiation intensity, as given by Eq. (12), over the
entire solid angle of 4 as
As an example, for the isotropic radiator ideal antenna, the radiation intensity U(,) will be independent
of the angles and , and the total radiated power will be
A term associated with the normalized power pattern is the beam solid angle
A defined as the solid
angle through which all the power from a radiating antenna would flow if the power per unit solid angle were
constant over that solid angle and equal to its maximum value (Fig. 8). This means that, for typical patterns,
the beam solid angle is approximately equal to the half-power beamwidth (HPBW), that is,
If the integration is done over the main lobe, the main-lobe solid angle,
M , results, and the difference of
A
M gives the minor-lobe solid angle. These definitions hold for patterns with clearly defined lobes. The
beam efficiency (BE) of an antenna is defined as the ratio
M /
A and is a measure of the amount of power in the
major lobe compared to the total power. A high beam efficiency means that most of the power is concentrated
in the major lobe and that minor lobes are minimized.
Directivity and Gain. A very important antenna parameter, which indicates how well an antenna
concentrates power into a limited solid angle, is its directivity D, defined as the ratio of the maximum radiation
intensity to the radiation intensity averaged over all directions. The average radiation intensity is calculated
15
since from Eq. (16), Prad /4 = U i . So, alternatively, the directivity of an antenna can be defined as the ratio of
its radiation intensity in a given direction (which usually is taken to be the direction of maximum radiation
intensity) to the radiation intensity of an isotropic source with the same total radiation intensity. Equation
16
Thus, the directivity of an antenna is equal to the solid angle of a sphere, which is 4 sr, divided by
the antenna beam solid angle
A . We can say that by this relation the value of directivity is derived from
the antenna pattern. It is obvious from this relation that the smaller the beam solid angle, the larger the
directivity, or, stated in a different way, an antenna that concentrates its power in a narrow main lobe has a
large directivity.
Obviously, the directivity of an isotropic antenna is unity. By definition an isotropic source radiates equally
in all directions. If we use Eq. (20),
A = 4, since U n (, ) = 1. This is the smallest directivity value that one
can attain. However, if we consider the directivity in a specified direction, for example D(, ), its value can be
smaller than unity.
As an example let us calculate the directivity of the very short dipole. We can calculate its normalized
radiated power using the electric or the magnetic field components, given in Table 1. Using the electric field E
for the far-field region, from Eq. (12) we have
and
Alternatively, we can work using power densities instead of power intensities. The power flowing in a
particular direction can be calculated using Eq. (3) and the electric and magnetic far-field components given in
Table 1:
17
By integrating over all angles the total power flowing outwards is seen to be
The directivity is the ratio of the maximum power density to the average power density. For the very short
dipole antenna, the maximum power density is in the = 90 direction (Fig. 2), and the average power density
is found by averaging the total power PT from Eq. (24) over a sphere of surface area 4 r2 . So
Thus, the directivity of a very short dipole is 1.5, which means that the maximum radiation intensity is
1.5 times the power of the isotropic radiator. This is often expressed in decibels:
Here, we use a lowercase letter for the absolute value and a capital letter for the logarithmic value of the
directivity, as is a common in the field of antennas and propagation.
The gain of an antenna is another basic property for its characterization. Gain is closely associated with
directivity, which is dependent upon the radiation patterns of an antenna. The gain is commonly defined as the
ratio of the maximum radiation intensity in a given direction to the maximum radiation intensity produced in
the same direction from a reference antenna with the same power input. Any convenient type of antenna may
be taken as the reference. Many times the type of the reference antenna is dictated by the application area, but
the most commonly used one is the isotropic radiator, the hypothetical lossless antenna with uniform radiation
intensity in all directions. So
where the radiation intensity of the reference antenna (isotropic radiator) is equal to the power in the input,
Pin , of the antenna divided by 4.
Real antennas are not lossless, which means that if they accept an input a power Pin , the radiated power
Prad generally will be less than Pin . The antenna efficiency k is defined as the ratio of these two powers:
where Rr is the radiation resistance of the antenna. Rr is defined as an equivalent resistance in which the same
current as that flowing at the antenna terminals would produce power equal to that produced by the antenna.
Rloss is the loss resistance, which allows for any heat loss due to the finite conductivity of the materials used to
construct the antenna or due to the dielectric structure of the antenna. So, for a real antenna with losses, its
radiation intensity at a given direction U(, ) will be
18
Thus, the gain of an antenna over a lossless isotropic radiator equals its directivity if the antenna efficiency
is k = 1, and it is less than the directivity if k < 1.
The values of gain range between zero and infinity, while those of directivity range between unity and
infinity. However, while the directivity can be computed from either theoretical considerations or measured
radiation patterns, the gain of an antenna is almost always determined by a direct comparison of measurement
against a reference, usually a standard-gain antenna.
Gain is expressed also in decibels:
where, as in Eq. (26), lowercase and capital letters mean absolute and logarithmic values, respectively. The
reference antenna used is sometimes declared in a subscript; for example, dBi means decibels over isotropic.
Polarization
Wave and antenna polarization. Polarization refers to the vector orientation of the radiated waves
in space. As is known, the direction of oscillation of an electric field is always perpendicular to the direction of
propagation. For an electromagnetic wave, if its electric field oscillation occurs only within a plane containing
the direction of propagation, it is called linearly polarized or plane-polarized. This is because the locus of
oscillation of the electric field vector within a plane perpendicular to the direction of propagation forms a
straight line. On the other hand, when the locus of the tip of an electric field vector forms an ellipse or a circle,
the electromagnetic wave is called an elliptically polarized or circularly polarized wave.
The decision to label polarization orientation according to the electric intensity is not as arbitrary as it
seems; its causes the direction of polarization to be the same as the direction of the antenna. Thus, vertical
antennas radiate vertically polarized waves, and horizontal antennas radiate horizontally polarized waves.
There has been a tendency, over the years, to transfer the label to the antenna itself. Thus people often refer
to antennas as vertically or horizontally polarized, whereas it is only their radiations that are so polarized.
It is a characteristic of antennas that the radiation they emit is polarized. These polarized waves are
deterministic, which means that the field quantities are definite functions of time and position. On the other
hand, other forms of radiation, for example light emitted by incoherent sources, such as the sun or light
bulbs, has a random arrangement of field vectors and is said to be randomly polarized or unpolarized. In
this case the field quantities are completely random and the components of the electric field are uncorrelated.
In many situations the waves may be partially polarized. In fact, this case can be seen as the most general
situation of wave polarization; a wave is partially polarized when it may be considered to be of two parts, one
completely polarized and the other completely unpolarized. Since we are mainly interested in waves radiated
from antennas, we consider only polarized waves.
Linear, Circular, and Elliptical Polarization. Consider a plane wave traveling in the positive z direction, with the electric field at all times in the x direction, as in Fig. 9(a). This wave is said to be linearly
19
Fig. 9. Polarization of a wave: (a) linear, (b) circular, and (c) elliptical.
polarized (in the x direction), and its electric field as a function of time and position can be described by
In general the electric field of a wave traveling in the z direction may have both an x and a y component,
as shown in Fig. 9(b, c). If the two components Ex and Ey are of equal amplitude, the total electric field at a
fixed value of z rotates as a function of time, with the tip of the vector forming a circular trace, and the wave is
said to be circularly polarized [Fig. 9(b)].
Generally, the wave consists of two electric field components, Ex and Ey , of different amplitude ratios
and relative phases. (Obviously, there are also magnetic fields, not shown in Fig. 9 to avoid confusion, with
amplitudes proportional to and in phase with Ex and Ey , but orthogonal to the corresponding electric field
vectors.) In this general situation, at a fixed value of z the resultant electric vector rotates as a function of time,
the tip of the vector describing an ellipse, which is called the polarization ellipse, and the wave is said to be
elliptically polarized [Fig. 9(c)]. The polarization ellipse may have any orientation, which is determined by its
tilt angle, as shown in Fig. 10; the ratio of the major to the minor axis of the polarization ellipse is called the
axial ratio (AR). Since the two cases of linear and circular polarization, can be seen as two particular cases
of elliptical polarization, we will analyze the latter. Thus, for a wave traveling in the positive z direction, the
electric field components in the x and y directions are
20
where Ex0 and Ey0 are the amplitudes in the x and y directions, respectively, and is the time-phase angle
between them. The total instantaneous vector field E is
If we define
21
which is the equation of an ellipse, the polarization ellipse shown in Fig. 10. The line segment OA is the
semimajor axis, and the line segment OB is the semiminor axis. The tilt angle of the ellipse is . The axial ratio
is
From this general case, the cases of linear and circular polarization can be found. Thus, if there is only
Ex (Ey0 = 0), the wave is linearly polarized in the x direction, and if there is only Ey (Ex0 = 0), the wave is
linearly polarized in the y direction. When both Ex and Ey exist, for linear polarization they must be in phase
or antiphase with each other. In general, the necessary condition for linear polarization is that the time-phase
difference between the two components must be a multiple of . If = 0, , 2, . . . and Ex0 = Ey0 , the wave is
linearly polarized but in a plane at an angle of /4 with respect to the x axis ( = /4). If the ratio of the
amplitudes Ex0 and Ey0 is different, then the tilt angle will also be different.
If Ex0 = Ey0 and = /2, the wave is circularly polarized. Generally, circular polarization can be achieved
only when the magnitudes of the two components are the same and the time-phase angle between them is an
odd multiple of /2.
Consider the case that = /2. Taking z = 0, from Eq. (33), (34), and (35) at t = 0 one has E = Ey0 ,
and one-quarter cycle later, at t = /2, one has E = Ex0 . Thus, at a fixed position (z = 0) the electric field
vector rotates with time, tracing a circle. The sense of rotation, also referred to as the sense of polarization, can
be defined by the sense of rotation of the wave as it is observed along the direction of propagation. Thus the
above wave rotates clockwise if it is observed looking towards the source (viewing the wave approaching) or
counterclockwise if it is observed looking away from the source (viewing the wave moving away). Thus, unless
the wave direction is specified, there is ambiguity. The most generally accepted notation is that of the IEEE,
by which the sense of rotation is always taken as that with the wave it traveling away from the observer. If the
rotation is clockwise, the wave is right-handed or clockwise circularly polarized (RH or CW). If the rotation is
counterclockwise, the wave is left-handed or counterclockwise circularly polarized (LH or CCW). Yet another
way to define the polarization is with the aid of helical-beam antennas. A right-handed helical-beam antenna
radiates (or receives) right-handed waves regardless of the position from which it is viewed, while a left-handed
one radiates right-handed waves.
Although linear and circular polarizations can be seen as special cases of elliptical, usually, in practice,
elliptical polarization refers to other than linear or circular. A wave is characterized as elliptically polarized
if the tip of its electric vector forms an ellipse. For a wave to be elliptically polarized, its electric field must
have two orthogonal linearly polarized components, Ex0 and Ey0 . If the two components are not of the same
magnitude, the time-phase angle between them must not be 0 or a multiple of , while in the case of equal
magnitude, the angle must not be an odd multiple of /2. Thus, a wave that is not linearly or circular polarized
is elliptically polarized. The sense of its rotation is determined according to the same rule as for circular
polarization. So a wave is right-handed or clockwise elliptically polarized (RH or CW) if the rotation of its
electric field is clockwise, and it is left-handed or counterclockwise elliptically polarized (LH or CCW) if the
electric field vector rotates counterclockwise.
In addition to the sense of rotation, elliptically polarized waves are characterized by their axial ratio
AR and their tilt angle . The tilt angle is used to identify the spatial orientation of the ellipse and can
22
Fig. 11. Polarization states of an electromagnetic wave represented with the aid of the Poincare sphere: (a) one octant of
the Poincare sphere with polarization states, (b) the full range of polarization states in rectangular projection.
be measured counterclockwise or clockwise from the reference direction (Fig. 10). If the electric field of an
elliptically polarized wave has two components of different magnitude with a time-phase angle between them
an odd multiple of /2, the polarization ellipse will not be tilted. Its position will be aligned with the principal
axes of the field components, so that the major axis of the ellipse will be aligned with the axis of the larger field
component and the minor axis with the smaller one.
The Poincare Sphere and Antenna Polarization Characteristics. The polarization of a wave can
be represented and visualized with the aid of a Poincare sphere. The polarization state is described by a
point on this sphere, whose longitude and latitude are related to parameters of the polarization ellipse. Each
point represents a unique polarization state. On the Poincare sphere the north pole represents left circular
polarization, the south pole right circular polarization, and the points along the equator linear polarization of
different tilt angles. All other points on the sphere represent elliptical polarization states. One octant of the
Poincare sphere with polarization states is shown in Figure 11(a), while the full range of polarization states is
shown in Figure 11(b), which presents a rectangular projection of the Poincare sphere.
The polarization state described by a point on Poincare sphere can be expressed in terms of:
(1) The longitude L and latitude of the point, which are related to the parameters of the polarization ellipse
by
where is the tilt angle with values 0 and = cot 1 ( AR) with values /4 +/4. The axial
ratio AR is negative and positive for right- and left-handed polarization respectively.
23
Fig. 12. One octant of the Poincare sphere, showing the relations of the angles , , , and that can be used to describe
a polarization state.
(2) The angle subtended by the great circle drawn from a reference point on the equator, and the angle between
the great circle and the equator:
where = tan 1 (Ey0 /Ex0 ) with 0 /2 and the time-phase difference between the components of the
electric field ( +).
All the above quantities , , , and , are interrelated by trigonometric formulae (4), and knowing , one
can determine , and vice versa. As a result, the polarization state can be described by either of these two
sets of angles. The geometric relation between these angles is shown in Fig. 12.
The polarization state of an antenna is defined as the polarization state of the wave radiated by the
antenna when it is transmitting. It is characterized by the axial ratio AR, the sense of rotation, and the tilt
angle, which identifies the spatial orientation of the ellipse. However, care is needed in the characterization
of the polarization of a receiving antenna. If the receiving antenna has a polarization that is different from
that of the incident wave, a polarization mismatch occurs. In this case the amount of power extracted by the
receiving antenna from the incident wave will be lower than the expected value, because of the polarization
loss. A figure of merit, which can be used as a measure of polarization mismatch, is the polarization loss factor
(PLF). It is defined as the cosine squared of the angle between the polarization states of the antenna in its
transmitting mode and the incoming wave. Another quantity that can be used to describe the relation between
the polarization characteristics of an antenna and an incoming wave is the polarization efficiency, also known
as loss factor or polarization mismatch. It is defined as the ratio of the power received by an antenna from a
given plane wave of arbitrary polarization to the power that would be received by the same antenna from a
plane wave of the same power flux density and direction of propagation, whose state of polarization has been
adjusted for maximum received power.
In general an antenna is designed for a specific polarization. This is the desired polarization and is called
the copolarization or normal polarization, while the undesired polarization, usually taken orthogonal to the
desired one, is known as the cross polarization or opposite polarization. The latter can be due to a change
24
of polarization characteristics during the propagation of waves, which is known as polarization rotation. In
general an actual antenna does not completely discriminate against a cross-polarized wave, due to engineering
and structural restrictions. The directivity pattern obtained over the entire direction on a representative
plane for cross polarization with respect to the maximum directivity for normal polarization is called antenna
cross-polarization discrimination, and it is an important factor in determining the antenna performance.
The polarization pattern gives the polarization characteristics of an antenna and is the spatial distribution
of the polarization of the electric field vector radiated by the antenna over its radiation sphere. The description
of the polarizations is accomplished by specifying reference lines, which are used to measure the tilt angles of
polarization ellipses, or the directions of polarization for the case of linear polarization.
where k2 = 2 0 0 , with 0 and 0 the magnetic permeability and electric permittivity of the air, respectively,
and is the radian frequency. r is the distance from any point in the source to the observation point. The fields
are then given by
and
25
where in 41a the scalar function V represents an arbitrary electric scalar potential, which is a function
of position. The fields radiated by antennas with finite dimensions are spherical waves, and in the far-field
region, the electric and magnetic field components are orthogonal to each other and form a TEM (transverse
electromagnetic mode) wave. Thus in the far field region, Eq. (41a) simplify to
and
So the problem becomes that of first evaluating the function A from the specified electric current density
on the antenna; then, using Eq. (42a), the E and H fields are evaluated and the radiation pattern extracted.
For example, for the case of a very short dipole the magnetic vector potential A is given by
which means that it is separable into two functions, each being a function of one variable only, and A is a
constant. Then Prad from Eq. (15) will be
26
If we take N equal divisions over the interval of the variable , and M equal divisions over the interval
2 of the variable , the two integrals can be calculated by a series approximation, respectively:
and
A computer program can easily evaluate the above equation. The directivity then is given by Eq. (19),
which is repeated here:
In the case that and variations are not separable, Prad can also be calculated by a computer program
using a slightly different expression,
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
27
MICHAEL CHRYSSOMALLIS
Democritus University of Thrace
CHRISTOS CHRISTODOULOU
The University of New Mexico
ANTENNA THEORY
ANTENNA THEORY
FUNDAMENTALS
Maxwells Equations
Antenna properties are analyzed with basic laws of physics.
These laws have been collected into a set of equations commonly referred to as Maxwells equations. (The presentation
in this section follows the textbook by Stutzman and Thiele
(1) where a more detailed treat may be found.) In the time
domain, these equations are
E =
B
B
M
t
(1)
H =
D
D
+T
t
(2)
D = (t)
(3)
If multiple frequencies are present, the solution to the equations may be found for each frequency separately and the results combined for the total solution. The linearity restriction
was only to ensure that the analysis would be properly restricted to a single frequency. For nonlinear media and some
complex problems, it is advantageous to solve the time-domain equations and obtain the frequency-domain form
through a Fourier (or Laplace) transform process. Computationally, the Fourier transform is usually obtained using a
fast Fourier transform (FFT).
Maxwells equations define relationships between the field
quantities, but do not explicitly provide information about the
media in which these fields exist. The material is characterized by three quantities: permittivity , permeability , and
conductivity . Sometimes the material conductivity is given
in inverse form as the resistivity 1/ . These quantities
relate the density and intensity quantities as well as the portion of the current due to conduction. Thus we have D E,
B H, and J E that lead to
and
B =m
(4)
The cross and dot derivatives are referred to as the curl and
divergence respectively. A supplementary equation that may
be deduced from the second and third equations is
T =
(t)
t
(5)
(6)
to obtain the phasor form of Maxwells equations and the continuity equation as
E = jB
BM
(7)
D +J
H = jD
(8)
D =
(9)
B =m
(10)
J = j
(11)
and
595
H M)
E = jH
(12)
E + Ji
H = ( + j)E
E = i
j
(13)
H = (m)
(15)
J i = ji
(16)
(14)
and
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
596
ANTENNA THEORY
(18)
H (rr ) =
1
E + e jkkr E e jkkr ], =
k [E
(19)
for the geometry of Fig. 1. This general form can be specialized to the far-field case for an antenna located near the origin by expanding R r r in a binomial series as
R = |rr r | = r2 2rr r + r2
(24)
r2
(rr r )2
r r
+
= r
+
r
2r
2r3
for r sufficiently small. Only the first term in this expansion,
r, needs to be retained for use in the denominator of Eq. (23).
However, more accuracy is needed for R in the exponential to
account for phase changes; so the second term of the expansion is used in the exponential:
R r r r
e jrr
4r
1
A
(20)
The corresponding electric-field intensity in simple media (using the Lorentz gauge for the potential) is given by
E=
1
[ 2A + A]
j
J (rr )e j r rr dv
(26)
1
k A )]
[ 2A k (k
j
(27)
1
k A]
[k
j
(28)
and
H
H=
Auxiliary Functions
Auxiliary functions are used to extend the solution of the
wave equation beyond the simple traveling plane-wave form.
If the magnetic sources are zero, then we may expand the
magnetic-flux density in terms of the curl of an auxiliary function, the magnetic vector potential A, or
(25)
The second term in Eq. (27) simply removes the radial portion A from the electric field.
Duality
Duality provides an extremely useful way to complete the development of the solution form as well as equating some
forms of antennas. To complete the previous set of equations
for the magnetic current and charge, we simply note that we
may change the variable definitions to obtain an identical
form of equations. Specifically, we replace
E H
(29a)
E
H E
(29b)
J M
(29c)
A F
(29d)
,
(29e)
(21)
Source volume v
A = J
J
( 2 + 2 )A
(22)
A (rr ) =
V
R = r r
P, field point
J (rr )
e j |rr rr |
dv
4|rr r |
(23)
Figure 1. Coordinates and geometry for solving radiation problems.
ANTENNA THEORY
Im
Im
Sources
PEC
Zmonopole =
Images
1
Z
2 dipole
(32)
Since the corresponding field is radiated into only a halfspace, the gain of the antenna defined as the peak power density in the far-field compared to the average power density
over the radiation region of the antenna is double for the
ground plane-fed antenna as
Gmonopole = 2Gdipole
and
, 1/
ANTENNA CHARACTERISTICS
E=
1
F
[ 2A + A ]
j
(30a)
H=
1
A
[ 2F + F ] +
j
(30b)
and
The alternate use of duality is to equate similar dual problems numerically. A classic problem is the relationship between the input impedance of a slot dipole and strip dipole.
The two structures are planar complements, each filling the
void of the other, and have input impedances which satisfy
2
4
(33)
(29f)
where is the intrinsic impedance of the medium. The solution forms for J and M may be combined for the total solution
as
ZslotZstrip =
597
(31)
There are a number of characteristics that describe an antenna as a device. Characteristics such as impedance and
gain are common to any electrical device. On the other hand,
a property such as radiation pattern is unique to the antenna.
In this section we discuss patterns and impedance. Gain is
discussed in the following section. We begin with a discussion
of reciprocity.
Reciprocity
Circuit Form. Reciprocity plays an important role in antenna theory and can be used to great advantage in calculations and measurements. Fortunately, antennas usually behave as reciprocal devices. This permits characterization of
the antenna as either a transmitting or receiving antenna.
For example, radiation patterns are often measured with the
test antenna operating in the receive mode. If the antenna is
reciprocal, the measured pattern is identical when the antenna is in either a transmit or a receive mode. In fact, the
following general statement applies: If nonreciprocal materials are not present in an antenna, its transmitting and receiving
properties are identical. A case where reciprocity may not hold
is when ferrite or active devices are included as a part of the
antenna.
Reciprocity is also helpful when examining the terminal
behavior of antennas. Consider two antennas, a and b shown
in Fig. 3. Although connected through the intervening medium and not by a direct connection path, we can view this
as a two-port network. Two port circuit analysis permits us
to write the following:
Antenna
a
Va = Zaa Ia + Zab Ib
(34a)
Vb = Zba Ia + Zbb Ib
(34b)
Antenna
b
598
ANTENNA THEORY
(35)
Therefore, mutual impedance Zba provides the coupling between a transmitting antenna and a receiving antenna. Reversing the situation by using antenna b as the transmitter
and antenna a as the receiver, leads to
Va = Zab Ib
(36)
J a E b M a H b ] dv
(41)
b E a = [J
V
If we go to the extreme of taking the test antenna to the surface of the problem antenna, then Eq. (41) becomes an equation that may be used for the solution of the currents on the
antenna.
The actual form of reaction is to suggest that the field b
reaction with current a is equal to the field a reaction with
current b, or
Ja E b M a H b ) dv = (J
Jb E a M b H a ) dv
(J
(42)
V
It can be seen from Eqs. (35) and (36) that if the applied currents are the same (Ia Ib I), then reciprocity is satisfied
(i.e., Va Vb) if
Zab = Zba
written symbolically as
b, a
= a, b
(43)
(37)
Antenna Impedance
1
I2
J E b ) dv
(J
(45)
(38)
J E b ) dv = (J
J E a ) dv = IVa
(J
(44)
(39)
That is, the input impedance equals the self impedance and
antenna a acts as if it is in free space.
Reaction Theorem. Reciprocity may also be stated in integral form by cross multiplying Maxwells equations by the opposite field for two separate problems, integrating and combining to obtain
E a H b E b H a ] dss = [(J
Ja E b M a H b )
[E
(40)
S
V
J b E a M b H a )] dv
(J
For antenna problems, the surface integral on the left of Eq.
(40) is taken to an infinite radius and the integral becomes
zero for finite antennas. This form is the typical field form
of reciprocity. This form also suggests constructing a second
problem that can be used as an auxiliary form to solve the
original problem. For instance, if Mb 0 and Jb is a point
dipole (or test antenna) of vector unit length lb located in the
ANTENNA THEORY
599
sin
H-plane
H
HP = 90
E-plane
x
(b)
(a)
y
z
(c)
(d)
Figure 4. Radiation from an ideal dipole. (a) Field components. (b) E-plane radiation pattern
polar plot. (c) H-plane radiation pattern polar plot. (d) Three-dimensional pattern plot.
field vectors follow from Eq. (30a); they are shown for an instant of time for which the fields are peak. The angular variation of E and H over the sphere is sin . An electric-field
probe antenna moved over the sphere surface and oriented
parallel to E will have an output proportional to sin ; see
Fig. 4(b). Any plane containing the z-axis has the same radiation pattern since there is no variation in the fields. A pattern taken in one of these planes is called an E-plane pattern
because it contains the electric vector. A pattern taken in a
plane perpendicular to an E-plane and cutting through the
test antenna (the xy-plane in this dipole case) is called an Hplane pattern because it contains the magnetic field H . The
E- and H-plane patterns, in general, are referred to as principal plane patterns. The E- and H-plane patterns for the ideal
dipole are shown in Fig. 4(b) and (c). These are polar plots in
which the distance from the origin to the curve is proportional
to the field intensity; they are often called polar patterns or
polar diagrams.
The complete pattern for the ideal dipole is shown in isometric view with a slice removed in Fig. 4(d). This solid polar
Az =
I(z )
e j R
dz
4R
(46)
600
ANTENNA THEORY
We can do this because in the far field r is very large compared to the antenna size, so r z z cos . In the phase
term R, we must be more accurate when computing the
distance from points along the line source to the observation
point. The integral Eq. (46) sums the contributions from all
the points along the line source. Although the amplitude of
waves due to each source point is essentially the same, the
phase can be different if the path length differences are a
sizable fraction of a wavelength. We, therefore, include the
first two terms of the series in Eq. (52) for the R in the numerator of Eq. (46) giving
z
P(0,y,z)
R
z z
R r z cos
x
Figure 5. Geometry used for field calculations of a line source along
the z-axis.
where has been used for typical radiation media. Due to the
symmetry of the source, we expect that the radiation fields
will not vary with . This lack of variation is because as the
observer moves around the source such that and z are constant, the appearance of the source remains the same; thus,
its radiation fields are also unchanged. Therefore, for simplicity we will confine the observation point to a fixed in the
yz-plane ( 90) as shown in Fig. 5. Then from Fig. 5 we
see that
y2 = y2 + z 2
(47)
z = r cos
(48)
y = r sin
(49)
y2 + (z z )2 = y2 + z2 2zz + (z )2
(50)
Substituting Eqs. (47) and (48) into Eq. (49), to put all field
point coordinates into the spherical coordinate system, gives
R = {r2 + [2rz cos + (z )2 ]}1/2
(51)
1
1
[2rz cos + (z )2 ] 3 [2rz cos + (z )2 ]2 +
2r
8r
(z )3 sin2 cos
(z )2 sin2
= r z cos +
+
+
2r
2r2
(52)
R = r+
(53)
(54)
Az =
I(z )
e j (rz cos )
e j r
dz =
4r
4r
I(z )e j z
cos
dz
(55)
j
k A)
k (k
(56)
(57)
Note that this result is the portion of the first term of Eq. (56)
which is transverse to r because jA j(Az sin
Az cos r). This form is an important general result for zdirected sources that is not restricted to line sources.
The radiation fields from a z-directed line source (any zdirected current source in general) are H and E, and are
found from Eqs. (27) and (28). The only remaining problem is
to calculate Az, which is given by Eq. (26) in general and by
Eq. (55) for z-directed line sources. Calculation of Az is the
focal point of linear antenna analysis. We shall return to this
topic after pausing to further examine the characteristics of
the far-field region.
The radiation field components given by Eqs. (27) and (28)
yield
E =
H = H
(58)
ANTENNA THEORY
factor in the field expressions; see Eq. (55). The ejr phase
factor indicates a traveling-wave propagating radially outward from the origin and the 1/r magnitude dependence leads
to constant power flow just as with the infinitesimal dipole.
In fact, the radiation fields of all antennas of finite extent
display this dependence with distance from the antenna.
Another way to view radiation field behavior is to note that
spherical waves appear to an observer in the far field to be a
plane wave. This local plane wave behavior occurs because
the radius of curvature of the spherical wave is so large that
the phase front is nearly planar over a local region.
If parallel lines (or rays) are drawn from each point on a
line current as shown in Fig. 6, the distance R to the far field
is geometrically related to r by Eq. (54), which was derived by
neglecting high order terms in the expression for R in Eq.
(52). The parallel ray assumption is exact only when the observation point is at infinity, but it is a good approximation
in the far field. Radiation calculations often start by assuming
parallel rays and then determining R for the phase by geometrical techniques. From the general source shown in Fig.
6, we see that
R = r r cos
(59)
(60)
P
R
J
dv
r
r
co
Figure 6. Parallel ray approximation for far-field calculations of radiation from a general source.
601
(D/2)2
=
2rff
16
(61)
2D2
(62)
The far-field region is r rff and rff is called the far-field distance, or Rayleigh distance. The far-field conditions are summarized as follows:
r>
2D2
(63a)
rD
(63b)
(63c)
2D2 / to
(64a)
(64b)
(64c)
602
ANTENNA THEORY
j r
e
A=
4r
Je
j r rr
dv
(65)
Jz e j r rr dv
(66)
e
4r
x = 0, y = 0, |z |
L
2
(73)
elsewhere
Az =
e j r
4r
L/2
L/2
Io e j z
cos
dz =
e j r
sin[(L/2) cos ]
Io L
4r
(L/2) cos
(74)
e j r
4r
j r
I
I(z ) = o
0
I(z )e j z
cos
dz
(68)
(70)
1
r E
(71)
This equation expresses the fact that in the far field the
directions of E and H are perpendicular to each other
and to the direction of propagation, and also that their
magnitudes are related by . For z-directed sources
H =
(75)
(67)
E = jA
A + j(r A )r = j(A + A )
sin[(L/2) cos ]
e j r
sin
4r
(L/2) cos
A
E = jA
(72)
E
E (max)
(76)
(77)
sin[(L/2) cos ]
(L/2) cos
(78)
(79)
ANTENNA THEORY
603
z
z
(b) Intermediate
(c) Endfire
(a) Broadside
(80)
for a z-directed current element. Actually this factor originates from Eq. (57) and can be interpreted as the projection
of the current element in the -direction. In other words, at
90 we see the maximum length of the current, whereas
at 0 or 180 we see the end view of an infinitesimal current which yields no radiation. The sin factor expresses the
fraction of the size of the current as seen from the observation
angle . On the other hand, the pattern factor f(, ) represents the integrated effect of radiation contributions from the
current distribution, which can be treated as being made up
of many current elements. The pattern value in a specific direction is then found by summing the parallel rays from each
current element to the far field with the magnitude and phase
of each included. The radiation integral of Eq. (65) sums the
far-field contributions from the current elements and when
normalized yields the pattern factor.
Antenna analysis is usually easier to understand by considering the antenna to be transmitting as we have here.
However, most antennas are reciprocal and thus their radiation properties are identical when used for reception; as discussed in the section on reciprocity.
For the z-directed uniform line source pattern Eq. (78) we
can identify the factors as
g( ) = sin
(81)
and
sin[(L/2) cos ]
(L/2) cos
(82)
(84)
(85)
(86)
f ( ) =
(83)
(88)
Radiation Pattern Parameters. A typical antenna power pattern is shown in Fig. 8 as a polar plot in linear units (rather
than decibels). It consists of several lobes. The main lobe (or
main beam or major lobe) is the lobe containing the direction
of maximum radiation. There is also usually a series of lobes
smaller than the main lobe. Any lobe other than the main
lobe is called a minor lobe. Minor lobes are composed of side
lobes and back lobes. Back lobes are directly opposite the
main lobe, or sometimes they are taken to be the lobes in
604
ANTENNA THEORY
Main lobe
1.0
0.5
Half-power beamwidth (HP)
Beamwidth between
first nulls (BWFN)
Minor
lobes
the half-space opposite the main lobe. The term side lobe is
sometimes reserved for those minor lobes near the main lobe,
but is most often taken to be synonymous with minor lobe;
we will use the latter convention.
The radiation from an antenna is represented mathematically through the radiation pattern function, F(, ) for field
and P(, ) for power. This angular distribution of radiation
is visualized through various graphical representations of the
pattern, which we discuss in this section. Graphical representations also are used to introduce definitions of pattern parameters that are commonly used to quantify radiation pattern characteristics.
A three-dimensional plot as in Fig. 4(d) gives a good overall
impression of the entire radiation pattern, but cannot convey
accurate quantitative information. Cuts through this pattern
in various planes are the most popular pattern plots. They
usually include the E- and H-plane patterns; see Figs. 4(b)
and (c). Pattern cuts are often given various fixed values,
leaving the pattern a function of alone; we will assume that
is the case here. Typically the side lobes are alternately positive and negative valued. In fact, a pattern in its most general
form may be complex-valued. Then we use the magnitude of
the field pattern F() or the power pattern P().
A measure of how well the power is concentrated into the
main lobe is the (relative) side lobe level, which is the ratio of
the pattern value of a side lobe peak to the pattern value of
the main lobe. The largest side lobe level for the whole pattern is the maximum (relative) side lobe level, frequently abbreviated as SLL. In decibels it is given by
F (SLL)
(89)
SLL = 20 log
F (max)
where F(max) is the maximum value of the pattern magnitude and F(SLL) is the pattern value of the maximum of the
highest side lobe magnitude. For a normalized pattern
F(max) 1.
The width of the main beam is quantified through halfpower beamwidth, HP, which is the angular separation of the
points where the main beam of the power pattern equals one-
(90)
where HP left and HP right are points to the left and right of the
main beam maximum for which the normalized power pattern has a value of one-half (see Fig. 8). On the field pattern
F() these points correspond to the value 1/2. For example,
the sin pattern of an ideal dipole has a value of 1/2 for
values of H-P left 135 and H-P right 45. Then HP 135
45 90. This is shown in Fig. 4(b). Note that the definition
of HP is the magnitude of the difference of the half-power
points and the assignment of left and right can be interchanged without changing HP. In three dimensions the
radiation pattern major lobe becomes a solid object and the
half-power contour is a continuous curve. If this curve is essentially elliptical, the pattern cuts that contain the major
and minor axes of the ellipse determine what the Institute
of Electrical and Electronics Engineers (IEEE) defines as the
principal half-power beamwidths.
Antennas are often referred to by the type of pattern they
produce. An isotropic antenna, which is hypothetical, radiates
equally in all directions giving a constant radiation pattern.
An omnidirectional antenna produces a pattern which is constant in one plane; the ideal dipole of Fig. 4 is an example.
The pattern shape resembles a doughnut. We often refer to
antennas as being broadside or endfire. A broadside antenna
is one for which the main beam maximum is in a direction
normal to the plane containing the antenna. An endfire antenna is one for which the main beam is in the plane containing the antenna. For a linear current on the z-axis, the
broadside direction is 90 and the endfire directions are
0 and 180. For example, an ideal dipole is a broadside antenna. For z-directed line sources several patterns are possible. Figure 7 illustrates a few f() patterns. The entire pattern (in three dimensions) is imagined by rotating the pattern
about the z-axis. The full pattern can then be generated from
the E-plane patterns shown. The broadside pattern of Fig.
7(a) is called fan beam. The full three dimensional endfire
pattern for Fig. 7(c) has a single lobe in the endfire direction.
This single lobe is referred to as a pencil beam. Note that the
sin element factor, which must multiply these patterns to
obtain the total pattern, will have a significant effect on the
endfire pattern. Intermediate scan angles are also possible,
as shown in Fig. 7(b).
ANTENNA PERFORMANCE MEASURES
Antennas are devices that are used in systems for communications or sensing. There are many parameters used to quantify the performance of the antenna as a device, which in turn
impacts on system performance. In this section we consider
the most important of these parameters when they are employed in their primary application area of communication
links, such as the simple communication link as shown in Fig.
9. We first discuss the basic properties of a receiving antenna.
Receiver
Transmitter
R
ANTENNA THEORY
IA
V
ZL
Incident
wave with
power
density, S
VA
ZL
ZA
Aem
(a)
(b)
1 |VA |2
1 |E i |2
=
(z)2
8 Rr
8 Rr
(92)
where Eq. (91) was used. The available power can also be
calculated by examining the incident wave. The power density
(Poynting vector magnitude) in the incoming wave is
S=
1
1 |E i |2
E H | =
|E
2
2
(93)
The available power is found using the maximum effective aperture Aem, which is the collecting area of the antenna. The
receiving antenna collects power from the incident wave in
proportion to its maximum effective aperture
PAm = SAem
1 |VA |2
PAm
3 2
1
8 Rr
=
=
(z)2 =
=
i 2
S
4
R
8
1 |E |
r
2
(95)
4 3 2
3
= 2
2
8
Ideal dipole
(96)
4
Aem
2
(97)
(91)
605
2
4
Isotropic antenna
(98)
(99)
which is also a general relationship. We can extract some interesting concepts from this relation. For a fixed wavelength
Aem and A, are inversely proportional; that is, as the maximum effective aperture increases (as a result of increasing its
physical size), the beam solid angle decreases, which means
power is more concentrated in angular space (i.e., directivity
goes up). For a fixed maximum effective aperture (i.e., antenna size), as wavelength decreases (frequency increases)
the beam solid angle also decreases, leading to increased directivity.
Directivity is more directly related to its definition through
this inverse dependence on beam solid angle as
D=
4
A
(100)
(94)
The maximum available power PAm will be realized if the antenna is directed for maximum response, is polarization
where
A =
|F (, )|2 d
(101)
606
ANTENNA THEORY
This directivity definition has a simple interpretation. Directivity is a measure of how much greater the power density at
a fixed distance is in a given direction than if all power were
radiated isotropically. This view is illustrated in Fig. 11. For
an isotropic antenna, as in Fig. 11(a), the beam solid angle is
4, and thus Eq. (100) gives a directivity of unity.
In practice antennas are not completely lossless. Earlier
we saw that power available at the terminals of a transmitting antenna was not all transformed into radiated power.
The power received by a receiving antenna is reduced to the
fraction er (radiation efficiency) from what it would be if the
antenna were lossless. This is represented by defining effective aperture
Ae = er Aem
S=
PA = SAe
S=
4
4
er Aem = 2 Ae
2
Ae = apAp
(105)
It is important to note that although we developed the general relationships of Eqs. (97), (99), and (104) for receiving
antennas, they apply to transmitting antennas as well. The
relationships are essential for communication system computations that we consider next.
Communication Links. We are now ready to completely describe the power transfer in the communication link of Fig. 9.
Um = DUave
Gt Pt Aer
4R2
Pr = Pt
Gt Gr 2
(4R)2
(b)
Figure 11. Illustration of directivity. (a) Radiation intensity distributed isotropically. (b) Radiation intensity from an actual antenna.
(108)
(109)
Aet Aer
R2 2
(110)
Uave
(107)
Uave
(a)
GtUave
GP
= t t2
R2
4R
Pr = SAer =
(104)
(106)
(103)
Uave
Pt
=
R2
4R2
(102)
and the available power with antenna losses included, analogous to Eq. (94), is
G = er D =
(111)
where
PD power delivered from the antenna
Pr power available from the receiving antenna
p polarization efficiency (or polarization mismatch factor), 0 p 1
q impedance mismatch factor, 0 q 1
ANTENNA THEORY
607
(112)
Temperature distribution
T (, )
(113)
4Um
= 4Um
Pt
TA
TA
TA
(a)
(b)
(114)
where Gt(dB) and Gr(dB) are the transmit and receive antenna gains in decibels, R(km) is the distance between the
transmitter and receiver in kilometers, and f(MHz) is the frequency in megahertz.
EIRP = Pt Gt
Rr
Power pattern
P (, )
(116)
The same radiation intensity could be obtained from a lossless isotropic antenna (with power gain Gi 1) if it had an
input power Pin equal to PtGt. In other words, to obtain the
same radiation intensity produced by the directional antenna
in its pattern maximum direction, an isotropic antenna would
have to have an input power Gt times greater. Effective isotropically radiated power is a frequently used parameter. For
example, FM radio stations often mention their effective radiated power when they sign off at night.
Antenna Noise Temperature and Radiometry
Receiving systems are vulnerable to noise and a major contribution is the receiving antenna, which collects noise from its
surrounding environment. In most situations a receiving antenna is surrounded by a complex environment as shown in
(117)
where
PNA available power due to antenna noise [W]
k Boltzmanns constant 1.38 1023 JK1
TA antenna temperature [K]
f receiver bandwidth [Hz]
Such noise is often referred to as Nyquist or Johnson noise
for system calculations. The system noise power PN is calculated using the total system noise temperature Tsys in place
of TA in Eq. (117) with Tsys TA Tr, where Tr is the receiver
noise temperature.
Antenna noise is important in several system applications
including communications and radiometry. Communication
systems are evaluated through carrier-to-noise ratio, which is
determined from the signal power and the system noise power
as
CNR =
PD
PN
(118)
608
ANTENNA THEORY
1
A
T (, )P(, ) d
(119)
To
A
P(, ) d =
0
To
= To
A A
(120)
(0.024)2
2
= 34 dB (70% aperture efficiency)
The received power from Eq. (114) is
Pr (dBm) = 20.8 + 34.2 + 34 20 log(38, 000) 20 log(12450)
32.44 = 116.9 dBW
(121)
PN = kTsys f
= 1.38 1023 125 20 106 = 3.45 1014
(122)
= 134.6 dBW
Thus the carrier to noise ratio from Eqs. (118) and (121) is
CNR(dB) = PD (dBW ) PN (dBW )
= 116.9 (134.6) = 17.7 dB
(123)
Antenna Bandwidth
Bandwidth is a measure of the range of operating frequencies
over which antenna performance is acceptable. Bandwidth is
computed in one of two ways. Let f U and f L be the upper and
lower frequencies of operation for which satisfactory performance is obtained. The center (or sometimes the design frequency) is denoted as f C. Then bandwidth as a percent of the
center frequency, Bp, is
Bp =
fU fL
100
fC
(124)
fU
fL
(125)
APERTURE ANTENNAS
WARREN L. STUTZMAN
WILLIAM A. DAVIS
Virginia Polytechnical Institute and
State University
609
ANTENNAS
563
ANTENNAS
HISTORY OF ANTENNAS
Marconis first experiments with transmitting electromagnetic waves in 1901, antennas have found several important
applications over the entire radio frequency range and, numerous designs of antennas now exist. Antennas are an integral part of our everyday lives, used for a multitude of purposes. An antenna is used to either transmit or receive
electromagnetic waves and it serves as a transducer converting guided waves into free space waves in the transmitting mode or vice versa in the receiving mode. All antennas
operate on the same basic principles of electromagnetic theory
formulated by James Clark Maxwell. Maxwell put forth his
unified theory of electricity and magnetism in 1873 (1) in his
famous book, A Treatise on Electricity and Magnetism, incorporating all previously known results on electricity and magnetism and expressing these mathematically through what we
refer to as Maxwells equations which hold over the entire
electromagnetic spectrum. His theory was met with much
skepticism and it wasnt until 1886 that Heinrich Hertz (2),
considered the father of radio, was able to validate this theory
with his experiments. The first radio system, at a wavelength
of 4 m, consisted of a /2 dipole (transmitting antenna) and a
resonant loop (receiving antenna) as shown in Fig. 1 (3). By
turning on the induction coil, sparks were induced across the
A
C
B
c
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
564
ANTENNAS
z
z
/2
ANTENNAS
tion techniques (29) are now used for a more accurate design
of these antennas.
The plane reflector is the simplest type of a reflector and
can be used to control the overall system radiation characteristics (21). The corner reflector has been investigated by
Kraus (30) and the 90 corner reflector is found to be the most
effective. The feeds for corner reflectors are generally dipoles
placed parallel to the vertex. These antennas can be analyzed
in a rather straightforward manner using the method of images. Among curved reflectors, the paraboloid is the most
commonly used. The paraboloid reflector shown in Fig. 3 is
formed by rotating a parabolic reflector about its axis. The
reflector transforms a spherical wave radiated from a feed at
its focus into a plane wave.
To avoid blockage caused by the feed placed at the focal
point in a front fed system, the feed is often off-set from the
axis (31). The Cassegrain reflector is a dual reflector system
using a paraboloid as the primary and a hyperboloid as the
secondary reflector with a feed along the axis of the paraboloid.
The Gregorian dual reflector antenna uses an ellipse as the
subreflector. The aperture efficiency in a Cassegrain antenna
can be improved by modifying the reflector surfaces (28). Most
paraboloidal reflectors use horn antennas (conical or pyramidal) for their feeds. With a paraboloidal reflector, beam scanning by feed displacement is limited. A spherical reflector provides greater scanning but requires more elaborate feed
design since it fails to focus an incident plane to a point.
Spherical reflectors can suffer from a loss in aperture and increased minor lobes due to blockage by the feed.
565
/(n-1)
Lens Antennas
At larger wavelengths, reflectors become impractical due to
the necessity of having large feed structures and tolerance
requirements. At low frequencies, the lens antenna is prohibitively heavy. Both lens antennas and parabolic reflectors use
free space as a feed network to excite a large aperture. The
feed of a lens remains out of the aperture and thus eliminates
aperture blockage and high side lobe levels. Dielectric lens
antennas are similar to optical lenses and the aperture of the
antenna is equal to the projection of the rim shape. Lenses
are divided into two categories: single-surface and dual-surface. In the single-surface lens refraction occurs only at one
surface. The other surface is an equiphase surface of the incident or emergent wave and the waves pass through normal to
the surface without refraction. Single-surface lenses convert
either cylindrical or spherical waves to plane waves. In a
dual-surface lens, refraction occurs at both lens surfaces. The
far field is determined by diffraction from the aperture. Dualsurface lenses allow more control of the pattern characteristics. Both surfaces are used for focusing and the second surface can be used to control the amplitude distribution in the
aperture plane. These simple lenses are many wavelengths
thick if their focal length and aperture are large compared to
a wavelength. The surface of the lens can be zoned by removing multiples of wavelengths from the thickness. The zoning
can be done either in the refracting or non-refracting surface
as shown in Fig. 4. The zoned lens is frequency sensitive and
can give rise to shadowing losses at the transition regions (5).
Artificial dielectric lenses in which particles such as metal
spheres, strips, disks, or rods are introduced in the dielectric
have been investigated by Kock (32). The size of the particles
has to be small compared to the wavelength. Metal plate
lenses using spaced conducting plates are used at microwave
frequencies. Since the index of refraction of a metal-plate medium depends on the ratio of the wavelength to the spacing
between the plates, these lenses are frequency sensitive. The
Luneberg lens is a spherically symmetric lens with an index
of refraction that varies as function of the radius. A plane
wave incident on this lens will be brought to a focus on the
opposite side. These lens antennas can be made using a series
of concentric spherical shells each with a constant dielectric.
Traveling Wave Antennas
566
ANTENNAS
Dummy coax.
cable
Driven
element
Reflector
Directors
Coax.
cable
Sectoral H-plane
Sectoral E-plane
Pyramidal
Conical
ANTENNAS
b2
a2
h2
b1
a1
h1
Figure 8. A coaxial fed (a) microstrip antenna and (b) stacked microstrip antenna.
567
568
ANTENNAS
Frequency
Range
330 kHz
Low frequency
(LF)
30300 kHz
3003000 kHz
High frequency
(HF)
330 MHz
30300 MHz
3001000 MHz
24 GHz
48 GHz
812 GHz
Ku
1218 GHz
K
Ka
Submillimeter
waves
1827 GHz
2740 GHz
12 GHz
Usage
Long distance telegraphy, navigation. Antennas are physically large but electrically
small. Propagation is accomplished using earths surface
and the ionosphere. Vertically polarized wave
Aeronautical navigation services, long distance communications, radio broadcasting. Vertical polarization
Regional broadcasting and
communication links, AM
radio
Communications, broadcasting, surveillance, CB radio
(26.96527.225 MHz). Ionospheric propagation. Vertical
and horizontal propagation
Surveillance, TV broadcasting
(5472 MHz), (7688 MHz),
and (174216 MHz), FM radio (88108 MHz). Wind profilers
Cellular communications, surveillance TV (470890 MHz)
Long range surveillance, remote sensing
Weather, traffic control,
tracking, hyperthermia
Weather detection, long range
tracking
Satellite communications, missile guidance, mapping
Satellite communications, altimetry, high resolution
mapping
Very high resolution mapping
Airport surveillance
Experimental stage
satellites operate at the L, S, or Ku band, but increasing demand for mobile telephony and high speed interactive data
exchange is pushing the antenna and satellite technology into
higher operational frequencies (47). Future satellites will be
equipped with antennas at both the Ku and the Ka bands.
This will lead to greater bandwidth availability. For example,
the ETS-VI (A Japanese satellite comparable to NASAs
TDRS), carries five antennas: an S-band phased array, a 0.4
m reflector for 43/38 GHz, for up and down links, an 0.8 m
reflector for 26/33 GHz, a 3.5 m reflector for 20 GHz, and a
2.5 m reflector for 30 GHz and 6/4 GHz. Figure 10 shows a
few typical antennas used on satellites. It is expected that
millions of households, worldwide, will have access to dual
Ku/Ka band dishes in the twenty-first century. These households will be able to enjoy hundreds of TV channels from
around the world. Moreover, low cost access to high speed,
voice, data, and video communications will be available to
more customers (48).
Personal/Mobile Communication Systems. The vehicular antennas used with mobile satellite communications constitute
the weak link of the system. If the antenna has high gain,
then tracking of the satellite becomes necessary. If the vehicle
antenna has low gain, the capacity of the communication system link is diminished. Moreover, hand-held telephone units
require ingenious design due to lack of real estate on the
portable device.
There is more emphasis now in enhancing antenna technologies for wireless communications, especially in cellular
communications, which will enhance the link performance
and reduce the undesirable visual impact of antenna towers.
Techniques that utilize smart antennas, fixed multiple
beams, and neural networks are now being utilized to increase the capacity of mobile communication systems,
whether it is land-based or satellite-based (49). It is anticipated that in the twenty-first century the wire will no
longer dictate where we must go to use the telephone, fax, email, or run a computer. This will lead to the design of more
compact and more sophisticated antennas.
Satellite
22500 miles
Satellite dish
Satellite dish
ANTENNAS
569
15.2 ft
C-band omni antenna
Dual subreflectors
47.1 ft
29.9 ft
Solar array
10.8-ft,
20-GHz
transmitting
antenna
Ka-band
command,
ranging, and
telemetry
antennas
Beam-forming
networks
7.2-ft, 30-GHz
receiving antenna
3.3-ft steerable antenna
Figure 10. Typical antennas on a satellite. (Courtesy, NASA Lewis Center)
Solar array
570
ANTENNAS
Transmitter
Duplexer
Low noise
amplifier
Display
Pulse
modulator
LO
Mixer
IF
amp.
Detector
Video
amp.
Receiver
Pr
Transmitter
Pt
max
min
ANTENNAS
571
Figure 13. A reflectivity map of the earth taken by NASAs Scatterometer. (Courtesy, NASA/
JPL)
still continues its original success using the new series of satellites called the Landsats. In 1985, British scientists noted
the ozone depletion over Antarctica. In 1986, US and French
satellites sensed the Chernobyl nuclear reactor explosion that
occurred in Ukraine. Landsat images from 1975 to 1986
proved to be very instrumental in determining the deforestation of the earth, especially in Brazil. In 1992, hurricane Andrew, the most costly natural disaster in the history of the
United States, with winds of 160 miles per hour, was detected
on time by very high resolution radar on satellites. Because
of the ability to detect the hurricane from a distance, on time,
through sophisticated antennas and imagery, the casualties
from this hurricane were low. In 1993, during the flooding of
the Mississippi River, antenna images were used to assist in
emergency planning, and locating threatened areas (56). In
1997, NASA, using antennas, managed to receive signals
from Mars and have the entire world observe the pathfinder
maneuver itself through the rocky Martian terrain.
10. R. S. Elliot, Antenna Theory and Design, New York: PrenticeHall, 1981.
BIBLIOGRAPHY
572
C. CHRISTODOULOU
P. F. WAHID
University of Central Florida
ANTENNAS,
ANTENNAS,
ANTENNAS,
ANTENNAS,
ANTENNAS,
APERTURE ANTENNAS
APERTURE ANTENNAS
Aperture antennas are most commonly used in the microwave- and millimeter-wave frequencies. There are a large
number of antenna types for which the radiated electromagnetic fields can be considered to emanate from a physical aperture. Antennas that fall into this category include several
types of reflectors: planar (flat plate) arrays, lenses, and
horns. The geometry of the aperture geometry may be square,
rectangular, circular, elliptical, or virtually any other shape.
Aperture antennas are very popular for aerospace applications because they can be flush mounted onto the spacecraft
or aircraft surface. Their opening can be covered with an electromagnetic (dielectric) window material or radome to protect
the antenna from environmental conditions (1). This installation will not disturb the aerodynamic profile of the vehicle,
which is of critical importance in high-speed applications.
In order to evaluate the distant (far-field) radiation patterns, it is necessary to know the internal currents that flow
on the radiating surfaces. However, these current distributions may not be exactly known and only approximate or experimental measurements can provide estimates for these
data. To expedite the process, it is necessary to have alternative methods to compute the radiation patterns of the aperture antennas. A technique based on the equivalence principle allows one to make a reasonable approximation to the
fields on, or in the vicinity of, the physical antenna structure
and subsequently to compute the radiation patterns.
Field equivalence, first introduced by Schelkunoff (2), is a
principle by which the actual sources on an antenna are replaced by equivalent sources on an external closed surface
that is physically outside of the antenna. The fictitious
sources are said to be equivalent within a region because they
produce the same fields within that region. Another key concept is Hugenss principle (3), which states that the equivalent source at each point on the external surface is a source
of a spherical wave. The secondary wave front can be con-
609
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
610
APERTURE ANTENNAS
field is integrated over two dimensions rather than one dimension (7). This double Fourier transform can also be applied to circular apertures and can be easily evaluated on a
PC.
For all aperture distributions, the following observations
are made (8):
1. A uniform amplitude distribution yields the maximum
directivity (nonuniform edge-enhanced distributions for
supergain being considered impractical), but at high
side-lobe levels.
2. Tapering the amplitude at the center, from a maximum
to a smaller value at the edges, will reduce the side-lobe
level compared with the uniform illumination, but it results in a larger (main-lobe) beam width and less directivity.
3. An inverse-taper distribution (amplitude depression at
the center) results in a smaller (main-lobe) beam width
but increases the side-lobe level and reduces the directivity when compared with the uniform illumination
case.
4. Depending on the aperture size in wavelengths and
phase error, there is a frequency (or wavelength) for
which the gain peaks, falling to smaller values as the
frequency is either raised or lowered.
Lastly, we consider aperture efficiencies. The aperture efficiency is defined as the ratio of the effective aperture area to
the physical aperture area. The beam efficiency is defined as
the ratio of the power in the main lobe to the total radiated
power. The maximum aperture efficiency occurs for a uniform
aperture distribution, but maximum beam efficiency occurs
for a highly tapered distribution. The aperture phase errors
are the primary limitation of the efficiency of the antenna.
HUYGENSS PRINCIPLE
The principle proposed by Christian Huygens (16291695) is
of fundamental importance in the development of the wave
theory. Huygenss principle states that, Each point on a primary wavefront serves as the source of spherical secondary
wavelets that advance with a speed and frequency equal to
those of the primary wave. The primary wavefront at some
later time is the envelope of these wavelets (9,10). This is
illustrated in Fig. 1 for spherical and plane waves modeled
Incoming
wave
Outgoing
wave
Spherical
wave
front
R0
ri
Incoming
wave
Plane-wave front
(a) Spherical
(b) Plane
Figure 1. Spherical and plane-wave fronts constructed with Huygens secondary waves.
(1)
For points near to the array, the integral does not simplify
but can be reduced to the form of Fresnel integrals.
The actual evaluation of this integral is best achieved on a
PC computer, which reduces the integral to the summation of
N Huygens sources:
E=
N
E0
i=1
e jkr i
ri
R0
2a2
5
10
15
(2)
611
20
20
E0
0
Distance along the y axis (cm)
(a) R0 = 2.5 cm
20
0
Distance along the y axis (cm)
(b) R0 = 5 cm
20
0
Distance along the y axis (cm)
(c) R0 = 15 cm
20
0
Distance along the y axis (cm)
(d) R0 = 20 cm
20
0
5
10
15
20
20
E=
0
5
10
15
20
20
APERTURE ANTENNAS
0
5
10
15
20
20
612
APERTURE ANTENNAS
In essence, Huygenss principle and the equivalence theorem shows how to replace actual sources by a set of equivalent sources spread over the surface S (13). The equivalence
principle is developed by considering a radiating source, electrically represented by current densities J1 and M1. Assume
that the source radiates fields E1 and H1 everywhere. We
would like to develop a method that will yield the fields outside the closed surface. To accomplish this, a closed surface S
is shown by the dashed lines that enclose the current densities J1 and M1. The volume inside S is denoted by V. The
primary task is to replace the original problem [Fig. 5(a)] by
an equivalent one that will yield the same fields E1 and H1
[Fig. 5(b)]. The formulation of the problem can be greatly
aided if the closed surface is judiciously chosen so that the
fields over most of the surface, if not the entire surface, are
known a priori.
The original sources J1 and M1 are removed, and we assume that there exists a field E and H inside V. For this field
to exist within V, it must satisfy the boundary conditions on
the tangential electrical and magnetic field components on
surface S. Thus on the imaginary surface S, there must exist
equivalent sources (14):
H1 H )
J S = n (H
(4)
n (E
E1 E )
M S = n
(5)
E 1, H 1
E 1, H 1
V1
S
E, H
J1
M1
JS = n (H1 H)
MS = n (E1 E)
(a) Horn
(b) Parabola
(c) Lens
APERTURE ANTENNAS
E(sin ) =
(6)
sin
(7)
E(sin )e jkx
E( ) =
L w /2
(8)
+L w /2
1
Lw
(9)
1
Lw
L w /2
e( j2 x/) sin dx
(10)
L w /2
E( ) =
L
sin
Lw sin
(11)
E(x) = cosn
0
d(sin )
sin
y
and
E(x) =
RECTANGULAR APERTURES
613
x
Lw
(12)
Lw
Figure 7. Coordinate system used to analyze a linear aperture of
length Lw.
614
APERTURE ANTENNAS
0.8
0.6
0.4
Uniform
cos
cos2
cos3
0.2
LW
2
0
Position along the x axis (cm)
Cp =
1
L w /2
L w /2
(13)
E (x) dx
E(x) = 1 +
To demonstrate the principles, we computed the antenna radiation pattern of a 1 meter long line-source antenna for
cosine0 (uniform), cosine1, and cosine2 distributions. The operating wavelength is 3 cm. The resulting patterns are shown
in Fig. 9. These data indicate that the more heavily tapered
illuminations result in decreased side-lobe levels, but at a
penalty of main beam peak gain.
Many distributions actually obtained in practice can be approximated by one of the simpler forms or by a combination
of simple forms. For example, a common linear aperture distribution is the cosine on a pedestal p:
x
(14)
Lw
x
Lw /2
(15)
x
Lw /2
(16)
for Lw /2 x 0, and
E(x) = 1
for 0 x Lw /2.
In practice, the rectangular aperture is probably the most
common microwave antenna. Because of its configuration, the
rectangular coordinate system is the most convenient system
to express the fields at the aperture. The most common and
convenient coordinate system used to analyze a rectangular
aperture is shown in Fig. 10. The aperture lies in the xy
plane and has a defined tangential aperture distribution E(x,
y). In keeping with the equivalence principle we shall assume
10
Uniform
cos
cos2
20
30
40
Figure 9. Radiation patterns of line
sources for three different aperture distributions (Lw 1 m, 3 cm).
50
LW
2
E(x) = p + (1 p) cos
(deg)
APERTURE ANTENNAS
Bw
Aw
y
E(, ) =
B w /2
B w /2
A w /2
(17)
A w /2
where
kx k sin cos
ky k sin sin
These are the x and y components of the propagation vector
k (20).
For many types of antennas, such as the rectangular horn,
the x and y functions are separable and may be expressed by
the form
E(x, y) = E(x)E(y)
sin kx
E(kx , ky ) =
Aw
2
Aw
kx
2
(18)
sin ky
Bw
2
Bw
ky
2
615
E(u) =
1
3
2
0
(20)
where
u=
Dw sin
(19)
P=
2r
Dw
2J1 (u)
u
(21)
CLS
Program to compute rectangular aperture radiation patterns
REDIM A(101, 101), patternx(200), patterny(200), intercardinal(200)
PI 3.14159265358#
Results stored in file Rect.dat
OPEN Rect.dat FOR OUTPUT AS #1
------ input data -----frequency 10
Aw 75: Bw 125
--------------------wavelength 30 / frequency
Awave Aw / wavelength: Bwave Bw / wavelength
Load aperture distribution: all dimensions in cm
FOR I 1 TO 101
FOR J 1 TO 101
X Aw * (51 - I) / 100: Y Bw * (51 - J) / 100
A(I, J) COS(PI * X / Aw) * COS(PI * Y / Bw)
Note: a(I, J) is the aperture distribution in the X and Y plane
NEXT J
NEXT I
Normalize power over aperture to that in a uniform distribution
power 0
FOR I 1 TO 101
FOR J 1 TO 101
power power A(I, J) 2
NEXT J
NEXT I
power power / 101 2 : power SQR(power)
Cp 1 / power
Cp is the desired normalization constant
FOR phi 0 TO 90 STEP 45
Note: Phi 0 results in principal X-plane pattern
Phi 45 results in intercardinal plane pattern
Phi 90 results in principal Y-plane pattern
K 0: G0DB 0
FOR theta 0 TO 10 STEP .1
PRINT Computing for theta; theta; Phi; phi
RE 0: IM 0
K K 1
Integration over aperture
FOR I 1 TO 101
FOR J 1 TO 101
X Aw * (51 - I) / 100: Y Bw * (51 - J) / 100
psi (2 * PI * X / wavelength) * SIN(theta * PI / 180) * COS(phi * PI / 180)
psi psi (2 * PI * Y / wavelength) * SIN(theta * PI / 180) * SIN(phi * PI / 180)
RE RE A(I, J) * COS(psi)
IM IM A(I, J) * SIN(psi)
NEXT J
NEXT I
TMM Cp * SQR(RE 2 IM 2) / 101 2
IF theta 0 THEN
G0DB 20 * LOG(TMM) / LOG(10): GV TMM
END IF
TOT 20 * LOG(TMM) / LOG(10)
IF phi 0 THEN patternx(K) TOT
IF phi 45 THEN intercardinal(K) TOT
IF phi 90 THEN patterny(K) TOT
NEXT theta
NEXT phi
K 0
FOR theta 0 TO 10 STEP .1
K K 1
PRINT #1, K, patternx(K), intercardinal(K), patterny(K)
PRINT Angle; theta; Ex; patternx(K); Exy; intercardinal(K): Ey; patterny(K)
NEXT theta
Print
PRINT Peak Gain G0 (dB); G0DB; G0(voltage); GV; G0(power); GV 2
CLOSE #1
END
Figure 11. Computer program to compute a radiation pattern from a rectangular aperture.
616
APERTURE ANTENNAS
617
x plane
Intercardinal
10
y plane
20
30
40
50
D
sin
2J1
E( ) =
Dw sin
Figure 12. Radiation patterns for a rectangular aperture (Aw 75 cm, Bw cm,
3 cm).
(22)
r
(23)
Dw
where 0 r Dw /2.
The cosine on a pedestal p distribution is here defined:
E(r) = p + (1 p) cos
r
(24)
D2
Distribution
Normalized
Half-Power
Beam Width
(deg)
HPbw/K
Normalized
Null-to-Null
Beam Width
(deg)
NULLbw*/K
Side-Lobe
Level
(dB):
SLL dB
Normalized
Side-Lobe
Angle (deg)
SLpos/K
50.67
114.67
13.26
82.00
1
2
3
4
5
68.67
82.67
95.33
106.00
116.67
172.00
229.33
286.67
344.00
402.00
23.00
31.46
39.29
46.74
53.93
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
68.67
64.67
62.00
59.33
58.00
56.00
54.67
54.00
52.67
52.00
50.67
172.00
162.00
152.67
144.67
138.00
132.67
127.67
123.33
120.00
117.33
114.67
73.34
114.59
Comments
Uniform
Cosine raised
to power n
Cosine on a
pedestal p
Triangular
n
n
n
n
n
p
p
p
p
p
p
p
p
p
p
p
Gain Relative
to Uniform
(dB)
G0 dB
Power Gain
Factor Relative
to Uniform
G0 power
Voltage Gain
Factor Relative
to Uniform
G0 volts
0.00
1.000
1.000
108.33
135.50
163.00
191.00
219.00
0.91
1.76
2.40
2.89
3.30
0.810
0.666
0.576
0.514
0.468
0.900
0.816
0.759
0.717
0.684
23.01
23.00
21.66
20.29
18.92
17.65
16.53
15.55
14.69
13.93
13.26
108.33
98.00
97.00
93.67
90.67
88.33
86.67
85.00
83.83
82.67
82.00
0.91
0.68
0.50
0.35
0.24
0.15
0.09
0.05
0.02
0.00
0.00
0.810
0.855
0.892
0.923
0.947
0.966
0.979
0.989
0.995
0.998
1.000
0.900
0.925
0.944
0.961
0.973
0.983
0.989
0.995
0.998
0.999
1.000
26.52
164.00
1.25
0.749
0.866
618
APERTURE ANTENNAS
efficiency M.
E(r) = 1
r
Dw /2
2
n
M =
(25)
M
A
(27)
(26)
D=
4
4
= 2 Ap
A
(28)
Ae
Ap
(29)
(30)
where M is the main beam solid angle (sr), A the total beam
solid angle (sr), and k0 the free space wavenumbers (k0
2/ ). It is important to recognize, then, that beam efficiency
and aperture efficiency are related to each other.
In terms of the radiated intensity E(, ) of a pencil beam
with boresight at ( 0, 0), the beam efficiency can be
defined by (14)
nB =
n n
2
0
(32)
where is the rms phase error over the aperture. It is assumed that the correlation intervals of the deviations are
greater than the wavelength. The controlling effect of the taper on the efficiencies (beam and aperture) tends to decrease
as the phase error increases. The efficiencies are also reduced
by the presence of the phase error.
The curves of Fig. 15 show that the beam efficiency tends
to increase with an increase in taper but the aperture efficiency decreases. Maximum aperture efficiency occurs for a
uniform aperture distribution, but maximum beam efficiency
occurs for a highly tapered distribution. In most cases a taper
is used that is intermediate between the two extremes.
r
x
y
Figure 13. Coordinate system used to analyze a circular aperture of
diameter Dw.
APERTURE SYNTHESIS
To simplify the discussion, we shall discuss a one-dimensional
line source or length Lw. Earlier in Eqs. (6) and (7) a Fourier-
APERTURE ANTENNAS
CLS
Program to compute rectangular aperture radiation patterns
REDIM A(101, 101), pattern(200)
PI 3.14159265358#
Results stored in file Circ.dat
OPEN Circ.data FOR OUTPUT AS #1
------ input data -----frequency 10: Dw 100
--------------------wavelength 30 / frequency: Dwave Dw / wavelength
Get reference power for uniformly illuminated circular array:
reference 0
FOR I 1 TO 101
FOR J 1 TO 101
X Dw * (51 - I) / 100: Y Dw * (51 - J) / 100
radius SQR(X 2 Y 2)
IF radius (Dw / 2) THEN
A(I, J) 0
ELSE
A(I, J) 1
END IF
reference reference A(I, J) 2
NEXT J
NEXT I
Load aperture distribution: all dimensions in cm
FOR I 1 TO 101
FOR J 1 TO 101
X Dw * (51 - I) / 100: Y Dw * (51 - J) / 100
radius SQR(X 2 Y 2)
IF radius (Dw / 2) THEN
A(I, J) 0
ELSE
A(I, J) COS(PI * radius / Dw)
Note: A(I, J) is aperture distribution in the X and Y plane
END IF
NEXT J
NEXT I
Normalize power over aperture to that in a uniform distribution
power 0
FOR I 1 TO 101
FOR J 1 TO 101
power power A(I, J) 2
NEXT J
NEXT I
power power / reference: power SQR(power)
Cp 1 / power
Cp is the desired normalization constant
phi 0
Note: Phi0 results in principal X-plane pattern
Phi45 results in intercardinal plane pattern
Phi90 results in principal Y-plane pattern
For a symmetrical circular distribution, all these are the same
K 0: G0DB 0
FOR theta 0 TO 10 STEP .1
PRINT Computing for theta; theta
RE 0: IM 0
K K 1
Integration over aperture
FOR I 1 TO 101
FOR J 1 TO 101
X Cw * (51 - I) / 100: Y Dw * (51 - J) / 100
psi (2 * PI * X / wavelength) * SIN(theta * PI / 180) * COS(phi * PI / 180)
psi psi (2 * PI * Y / wavelength) * SIN(theta * PI / 180) * SIN(phi * PI / 180)
RE RE A(I, J) * COS(psi)
IM IM A(I, J) * SIN(psi)
NEXT J
Figure 14. Computer program to compute the radiation pattern from a circular aperture.
619
620
APERTURE ANTENNAS
NEXT I
TMM Cp * SQR(RE 2 IM 2) / reference
IF TMM 0 THEN TMM 10 -6
IF theta 0 THEN
G0DB 20 * LOG(TMM) / LOG(10): GV TMM
END IF
pattern(K) 20 * LOG(TMM) / LOG(10)
NEXT theta
K 0
FOR theta 0 TO 10 STEP .1
K K 1
PRINT #1, theta, pattern(K)
PRINT Angl; theta; E ; pattern(K)
NEXT theta
Print
PRINT Peak Gain G0 (dB); G0DB; G0(voltage); GV; G0(power); GV
CLOSE #1
END
Figure 14 (Continued)
transform pair was defined for a line source relating the aperture distribution and the far-field radiation pattern, and vice
versa. In the synthesis process, we wish to determine the aperture distribution for a desired radiation pattern. To do this
we first express the illumination function as a sum of N uniform distributions:
E(x) =
N
cm e j m x
E( ) =
Lw
N
L w m=1
sin + m )x
dx
E( ) =
(33)
N
sin (k sin + m )
Cm
m=1
m=1
cm e j(k
(k sin + m )
Lw
2
(34)
(35)
Lw
2
Distribution
Normalized
Half-Power
Beam Width
(deg)
HPbw/K
Normalized
Null-to-Null
Beam Width
(deg)
NULLbw*/K
Side-Lobe
Level
(dB):
SLL dB
Normalized
Side-Lobe
Angle (deg)
SLpos/K
59.33
140.00
17.66
93.67
1
2
3
4
5
74.67
88.00
99.33
110.00
120.00
194.67
250.00
306.67
362.67
420.00
26.07
33.90
41.34
48.51
55.50
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
74.67
70.67
68.67
66.00
64.67
63.33
62.00
61.33
60.67
60.00
59.33
194.67
183.33
174.00
166.00
159.60
154.67
150.67
147.33
144.00
142.00
140.00
59.33
72.67
84.67
94.67
104
140
187.33
232.67
277.2
320.33
Comments
Uniform
Cosine raised
to power n
Cosine on a
pedestal p
Parabolic
raised to
power n
n
n
n
n
n
p
p
p
p
p
p
p
p
p
p
p
n
n
n
n
n
0
1
2
3
4
Gain Relative
to Uniform
(dB)
G0 dB
Power Gain
Factor Relative
to Uniform
G0 power
Voltage Gain
Factor Relative
to Uniform
G0 volts
0.00
1.000
1.000
119.33
145.50
173.00
200.30
228.17
1.42
2.89
4.04
4.96
5.73
0.721
0.514
0.394
0.319
0.267
0.849
0.717
0.628
0.564
0.517
26.07
25.61
24.44
23.12
21.91
20.85
19.95
19.18
18.52
17.96
17.66
119.33
112.67
107.83
104.17
101.33
99.47
97.83
96.47
95.27
94.57
93.67
1.42
0.98
0.66
0.43
0.27
0.17
0.09
0.05
0.02
0.00
0.00
1.000
0.799
0.859
0.905
0.9388
0.963
0.9789
0.9895
0.9958
0.9991
1.000
1.000
0.894
0.927
0.951
0.9689
0.981
0.9894
0.9947
0.9979
0.9995
1.000
17.66
24.64
30.61
35.96
40.91
93.67
116.33
138.67
160.17
181.33
0.00
1.244
2.547
3.585
4.432
1.00
0.701
0.556
0.438
0.36
1.00
0.866
0.746
0.662
0.6
APERTURE ANTENNAS
621
2. S. A. Schelkunoff, Some equivalence theorems of electromagnetics and their application to radiation problems, Bell System Tech.
J., 15: 92112, 1936.
Reading List
100
Efficiency (%)
80
60
40
20
0
(Full taper)
Beam efficiency
Aperture efficiency
Taper (dB)
0
(No taper)
beam, and there are N beams. The coefficients may be obtained manually by estimating the number of independent
beams and their relative magnitudes and positions to approximate the desired radiation pattern. Alternatively they may
be obtained mathematically via a Fourier-series representation. Also, the results may be extended by the reader to a twodimensional aperture.
The preceding equations form the basis for Woodwards aperture synthesis technique (6,29) that enables the aperture
illumination required to produce a given beam shape to be
approximated. A new array antenna synthesis method, called
the virtual array synthesis method, was recently published by
Vaskelainen (23). In this method, the excitation values of a
virtual array are synthesized using some known synthesis
method. The geometry of the virtual array can be chosen so
that there will be a suitable synthesis method for that geometry, and the synthesis of the virtual array can be done accurately enough. In the synthesis method, the excitation values
of the virtual array are transformed into the excitation values
of the actual array geometry. Matrix operations are simple
and large arrays can be easily synthesized. Further references on recent synthesis techniques appear in Refs. 24 to 28.
BIBLIOGRAPHY
1. D. J. Kozakoff, Analysis of Radome Enclosed Antennas, Norwood,
MA: Artech House, 1997.
622
C. L. Dolph, A current distribution for broadside arrays which optimizes the relationship between the beamwidth and the sidelobe
level, Proc. IRE, 34: 335348, 1946.
R. C. Hansen, A one parameter circular aperture with narrow beamwidth and low sidelobe levels, IEEE Trans. Antennas Propag., AP24: 477480, 1976.
R. C. Hansen, Microwave Scanning Antennas, New York: Academic
Press, 1966.
R. C. Hansen, Tables of Taylor distributions for circular aperture antennas, IRE Trans. Antennas Propag. AP-8: 2326, 1960.
R. F. Hyneman, A technique for the synthesis of line source antenna
patterns having specified sidelobe behavior, IEEE Trans. Antennas
Propag., AP-16: 430435, 1968.
H. Jasik, Fundamentals of Antennas, in R. C. Johnson (ed.), Antenna
Engineering Handbook, 3rd ed., New York: McGraw-Hill, 1993.
J. D. Kraus, Radio Astronomy, 2nd ed., New York: Cygnus-Quasar,
1986.
R. T. Nash, Beam efficiency limitations of large antennas, IEEE
Trans. Antennas Propag., AP-12: 918923, 1964.
J. F. Ramsay, Fourier transforms in aerial theory, Marconi Rev., 10:
17, 41, 81, 157, 1947.
J. F. Ramsay, Fourier transforms in aerial theory, Marconi Rev., 9:
139, 1946.
D. R. Rhodes, Synthesis of Planar Antenna Sources, London: Oxford
Univ. Press, 1974.
J. Ruze, Physical limitations of antennas, MIT Research Laboratory
Electronic Technical Report No. 248, 1952.
S. A. Schelkunoff, Some equivalence theorences of electromagnetics
and their application to radiation problems, Bell Syst. Tech. J., 15:
92112, 1936.
H. E. Shanks, A geometrical optics method of pattern synthesis for
linear arrays, IRE Trans. Antennas Propag., AP-8: 485490, 1960.
M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York:
McGraw-Hill, 1980.
J. C. Slater and N. H. Frank, Introduction to Theoretical Physics, New
York: McGraw-Hill, 1933.
T. T. Taylor, Design of circular apertures for narrow beamwidth and
low sidelobe levels, IEEE Trans. Antennas Propag., AP-8: 1722,
1960.
T. T. Taylor, Design of line-source antennas for narrow beamwidth
and low sidelobes, IRE Trans. Antennas Propag., AP-3: 1628,
1955.
I. Wolf, Determination of the radiating system which will produce a
specified directional characteristic, Trans. IRE, 25: 630643, 1937.
P. M. Woodward, A method of calculating the field over a plane aperture required to produce a given polar diagram, IEE J., Part III
A, 93: 15541558, 1947.
P. M. Woodward and J. D. Lawson, The theoretical precision with
which an arbitrary radiation pattern may be obtained from a
source of a finite size, IEE J., Part III, 95: 363370, 1948.
DENNIS KOZAKOFF
Millimeter Wave Technology
181
182
Abbreviation
ULF
ELF
VLF
LF
MF
HF
VHF
UHF
SHF
EHF
Frequency Range
3
3
3
30
300
3
30
300
3
30
300
Hz
Hz to 3 kHz
kHz to 30 kHz
kHz to 300 kHz
kHz to 3 MHz
MHz to 30 MHz
MHz to 300 MHz
MHz to 3 GHz
GHz to 30 GHz
GHz to 300 GHz
GHz to 1 THz
Mm
Mm to 100 km
km to 10 km
km to 1 km
km to 100 m
m to 10 m
m to 1 m
m to 10 cm
cm to 1 cm
cm to 1 mm
mm to 300 m
Source: The New IEEE Standard Dictionary of Electrical and Electronics Terms, 5th ed. The Institute of Electrical and Electronics
Engineers, 1993, Ref. 1, with modifications by the author. By permission of the IEEE.
cation by radio. Although radio waves also propagate in waveguides, cables, and transmission lines, the term radiowave
propagation is used generally, and specifically in this article,
for signal travel between a source and a receiver without
guidance by such artificial devices. Examples of such systems
are radio and television broadcasting, radio point-to-point
communication, radar, radio navigation, and remote sensing.
The concept of radiowave propagation allows dividing the
total system problem into three separate parts: the transmitter or signal source, the receiver, and propagation. Thus propagation is defined as what happens between the source and
the receiver. For this concept to be valid, the distances must
be such that far-field criteria for the antennas are satisfied;
this allows separating antenna effects from propagation effects. In radar two propagation paths are involved: from the
transmitter to the target and from the target to the receiver.
Much has been learned about radiowave propagation since
Hertz first demonstrated wireless transmission from one
room to another. Now radiowave propagation is considered a
moderately mature science, yet new understanding continues
to emerge. This article presents an overview of radiowave
propagation concepts. More detailed information will be found
in articles on specific topics in radiowave propagation.
Frequency Range
3
30
300
1
2
4
8
12
18
27
40
75
110
300
MHz to 30 MHz
MHz to 300 MHz
MHz to 1 GHz
GHz to 2 GHz
GHz to 4 GHz
GHz to 8 GHz
GHz to 12 GHz
GHz to 18 GHz
GHz to 27 GHz
GHz to 40 GHz
GHz to 75 GHz
GHz to 110 GHz
GHz to 300 GHz
GHz to 3 THz
Approx. Free-Space
Wavelength Range
100 m to 10 m
10 m to 1 m
1 m to 30 cm
30 cm to 15 cm
15 cm to 7.5 cm
7.5 cm to 3.75 cm
3.75 cm to 2.5 cm
2.5 cm to 1.67 cm
1.67 cm to 1.11 cm
1.11 cm to 7.5 mm
7.5 mm to 4 mm
4 mm to 2.73 mm
2.73 mm to 1 mm
1 mm to 100 m
Source: The New IEEE Standard Dictionary of Electrical and Electronics Terms,
5th ed. The Institute of Electrical and Electronics Engineers, 1993, Ref. 2,
with modifications by the author. By permission of the IEEE.
a
The mm designation is sometimes used for the entire frequency range 40 GHz
to 300 GHz when general information for this range is to be conveyed.
183
184
When the emitter and receiver are not within plain view of
each other at VHF and higher frequencies, diffraction is an
appropriate propagation mechanism. When the intervening
terrain is hilly or mountainous, the diffracting obstacles are
often modeled as cylinders or even as vertical half planes
(knife-edge diffraction). Diffraction by a conducting sphere
is the most common model for relatively flat terrain. More
detailed computational models are becoming increasingly
available as a result of increased computer power, more theoretical insights, and better algorithms.
Ray theory requires that the medium vary slowly with respect
to wavelength. Thus it fails for radiowave propagation at frequencies lower than about 100 kHz because the corresponding
wavelengths no longer meet the requirement that they be
much smaller than the earth-ionosphere distance. Instead, at
VLF and ELF the earth-ionospheric space is modeled as a
concentrically spherical waveguide with the ground as the
lower boundary and the ionosphere as the upper. In the ELF
and lower VLF bands only one, or at most a few, modes need
to be considered. The number of modes and therefore the computational difficulty increases with frequency. Calculations in
the LF frequency band are especially laborious, and a combination of ray and mode theory is often used.
Ground Wave
Ionospheric Scatter
Terrain Diffraction
Meteor Scatter
Tropospheric Scatter
The troposphere is never truly homogeneous, as common experience with weather indicates. In fact, variations over a
wide range of dimensional scales are ever present. These variations can be used to advantage when very reliable wireless
point-to-point communications are needed over a path of hundreds of kilometers. When very strong signals are beamed at
an atmospheric region within the line of sight of both stations, the relatively small signal scattered out of the beam
toward the receiving station may, nevertheless, be sufficient
to enable communication at rates on the order of 5 megabits
per second over 150 km to 400 km distances at UHF. This
technique is useful for communication with communities in
arctic regions, where other forms of communication (e.g., satellite communications, ionospheric reflection) are less reliable.
Ionospheric Reflection
In the upper LF, MF, and HF bands, signals give the appearance of traveling in rays that are reflected by the ionosphere
above and by the ground below, resulting in a series of hops.
Actually, the rays are bent rather than sharply reflected in
the ionosphere, but the effect is essentially the same. Signal
transmission by this means is very efficient, and great distances are spanned with modest power and equipment. For
these reasons the short-wave bands, as they are often called
popularly, are utilized for broadcasting, point-to-point communications, and amateur use, and portions of this band of
185
System Loss
System loss (Ls or As) is the ratio of the RF power input to the
transmitting antenna terminals and the resultant RF power
available at the receiving antenna terminals. It is expressed
in decibels by
Ls = 10 log(pt /pa ) = Pt Pa
(1)
where pt denotes the RF power input to the transmitter antenna terminals, pa the available power at the receiving antenna terminals, both measured in the same units, and Pt,
Pa are the corresponding power relative to a common reference level (e.g., 1 W) in decibels. Because the system loss is
referred to the antenna terminals, it excludes all losses in
feeder lines but includes all losses associated with the antenna, for example, antenna-grounding losses, dielectric
losses in the antenna, antenna loading-coil losses, and antenna-terminating resistor losses, if applicable.
Nonatmospheric Propagation
Transmission Loss
The propagation mechanisms discussed so far are those commonly encountered with electromagnetic signal propagation
through space or the earths atmosphere. A different environment prevails when signals propagate through the ocean,
through the earths crust, or through some planetary atmospheres. In the ocean, only signals at ELF or in the visiblelight frequency band penetrate to substantial depths. Propagation to substantial depths through the ground also presents
difficulties. Nevertheless, radar systems for locating objects,
such as water and gas lines at typical shallow depths, are
quite successful.
TRANSMISSION LOSS
The most significant question relating to propagation for a
radar or telecommunication system is the adequacy of the received signal level. The ratio, usually expressed in decibels
(dB), between the power at some point in the transmitting
system and a corresponding point in the receiving system is
called a loss. Although time delay or phase shift may also be
important parameters in some systems, loss is a universal
and primary consideration because it relates directly to signal
strength. In many cases, the loss varies randomly in time or
with location. Then statistical parameters relating to the loss
are often the quantities of interest, for example, the median
loss.
Several types of losses are defined; see Fig. 1. The usual
symbol for loss is L, although A (for attenuation) is also used,
and subscripts denote the specific loss under consideration.
The loss definitions below follow, with some emendations,
those in The concept of transmission loss for radio links,
Recommendation ITU-R P.341-4, of the International Telecommunication Union (4).
Total Loss
The total loss (Ll or Al) of a link is the ratio between the power
supplied by the transmitter and the power supplied to the
corresponding receiver under real installation, propagation,
and operational conditions. Note: it is necessary to specify the
points at which these powers are determined.
4d
(2)
186
Isotropic
antenna
Isotropic
antenna
Free space
Propagation
medium,
etc.
Basic transmission loss, Lb
Lb = Lbf + Lm
Transmitting
antenna
Gr
Gt
Receiving
antenna
Transmission loss, L
L = Lb Gt Gr
Transmitting
antenna
losses
Ltc
Lrc
Receiving
antenna
losses
Transmitter
Receiver
Filters,
feeder,etc.
Filters,
feeder,etc.
Total loss, Lb
(reference points should be specified )
System loss, Ll
Ls = L + L tc + + L tc = Pt Pa
(3)
Loss relative to free space may be divided into losses of different types, for example, extinction loss, diffraction loss, waveinterference loss, reflection loss, polarization-coupling loss,
aperture-to-medium coupling loss. Because these losses are
directly connected with propagation mechanisms, the end result of propagation studies is frequently presented in the form
of loss relative to free space.
Polarization-Coupling Loss. In an isotropic medium, polarization-coupling loss occurs when the receiving and transmitting antenna polarizations are not matched for optimum
187
BACKSCATTER
The scattering of waves from an object is a diffraction process
that transforms the incident wave into the waves propagating
from the object in all possible directions. The backscatter, or
backcattering, is the scattering of waves back toward the
source of the incident wave. This process substantially depends on material properties of the scattering object, its
shape, size, and spatial orientation relative to the incident
wave, as well as on the frequency and polarization of the incident wave. One distinguishes three frequency regions with
different physical properties of scattered waves. They are
quasi-static, resonance, and quasi-optical regions.
QUASI-STATIC SCATTERING
In the quasi-static region (sometimes called the Rayleigh region), the wavelength of the incident wave is much greater
than the maximum linear dimension l of the scattering object
( l). At a certain time t, the scattered field at small distances (r ) from the object is approximately the static field
created by dipoles and multipoles induced by the incident
wave in the scattering object at the same time t. Far from the
object, the scattered field is an outgoing spherical wave. Its
average power flux density (over one period of oscillations)
relative to that of the incident wave, or radar cross-section
(RCS), is determined by the objects volume rather than the
shape of the object, and it is proportional to 4. Specifically,
this dependence explains the blue color of the cloudless sky
during the day. This color is due to the predominance of blue
light scattered by small inhomogeneities of the air caused by
fluctuations of its density. The violet light, which is shorter
in wavelength, undergoes stronger relative scattering and as
a result suffers higher attenuation while propagating through
the atmosphere. Analytical expressions for quasi-static RCS
of some scattering objects can be found in (13). For example,
the axial RCS of perfectly conducting bodies of revolution is
determined by the following approximation (p. 145 of Ref. 1):
4
e
k 4V 2 1 +
64 4 6
k a
9
RESONANCE SCATTERING
In the resonance frequency region, linear dimensions of scattering objects are comparable to the wavelength of the incident wave. Eigen-oscillations excited by the incident wave in
the scattering object can substantially influence the scattering property. Frequencies of these oscillations are complex
quantities. Their imaginary parts determine both the internal
(thermal) losses inside the object and the external losses that
are due to radiation into the surrounding medium. A major
contribution to RCS is given by the radiation of those eigenoscillations, whose eigenfrequency real part and polarization
are close to the incident wave frequency and polarization. If
the quality factor of these oscillations is quite large, the amplitude and the intensity of the scattered wave sharply increase as the frequency of the incident wave approaches the
real part of the frequency of eigen-oscillations. This phenomenon is referred to as resonance scattering. It reveals itself,
for example, in scatterings from thin metallic half-wavelength
long wires and narrow strips (pp. 293303 of Ref. 2). These
scatterers are used, in particular, to create a chaff clutter for
radars. Analytical and numerical data for scattering by thin
wires are collected in chapter 12 of Ref. 3. Recent analytical
results for resistive wires are presented in Ref. 8.
The resonance scattering from wires can also be explained
as being due to the constructive interference of multiple current waves arising from the wire ends. This process is investigated in detail in Ref. 9. The total current generated by the
incident plane wave in the perfectly conducting wire (Fig. 1)
equals
2
J(z) = J0 (z) +
(1)
(3)
n=1
power of the wave number k 2/ . The expansion coefficients are found from the solution of the recursive system of
boundary value problems in potential theory (pp. 848856 of
Ref. 4). In practice it is possible to find only a few first coefficients. Direct numerical methods are efficient tools for the solution of quasi-static scattering problems (57).
(2)
for the circular disk with radius a. Table 8.2 in Ref. 2 (Vol. 2,
pp. 558561) contains explicit expressions for RCS found in
this manner for a variety of bodies of revolution. The first
term in this table (Eq. 8.1-87a on p. 558) contains a misprint.
The letter b should be replaced by h.
In this frequency region, the scattered field can be expressed in terms of a convergent series in positive integer
Jn (l),
Jn+1
(l)
J1 (l) = J0 (l)
Jn+ (l),
(4)
n = 1, 2, 3, . . .
(5)
175
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
176
BACKSCATTER
z
z=l
z=+l
J2n
(z) = J1+ (l)[ (kL)eikL]2n2[k(l z)]eik(lz)
(6)
J2n+1
(z) = J1 (l)[ (kL)eikL]2n1[k(l z)]eik(lz)
i
2 ln
ka
(kz) =
2ikz
E(2kz)e2ikz
ln
q
A
(2l)2
(7)
1.0
= 0.1
where
0.1
and
E(x) =
x
eit
dt = Ci(x) + iSi(x)
t
(8)
= 0.05
0.01
Functions Ci(x) and Si(x) are the well-tabulated cosine and
sine integrals, respectively. For small arguments (x 1),
function E(x) reduces to E(x) ln(x) ln(i) O(x) and ensures the equality (0) 1. Equation (6) shows that all multiple edge waves starting with secondary waves (n 2, 3, 4,
. . .) are expressed approximately by the same function (x).
As a result, the substitution of expressions (6) into Eq. (3)
leads to the geometric series
[Jn+ (z) + Jn (z)] = f (k, z, l, a)
{[ (kL)eikL ]2 }m
n=2
m=0
f (k, z, l, a)
=
D
(9)
0
0.001
0.5
1.0
1.5
2.0
2.5
3.0
3
2
5
2
2l
0.0001
Figure 2. Integral cross section of thin wires (from Ref. 11). This
quantity has the maximum (resonance) values for wires with the total
length L 2l (2n 1) /2, n 1, 2, 3, . . .. Along such wires from
one edge to another, each multiple edge wave acquires the phase shift
of (2n 1). Due to reflection at the edge, it acquires an additional
phase shift of . As a result, this wave becomes equi-phased with all
other multiple edge waves. This leads to the resonance behavior for
the current and scattered field.
BACKSCATTER
1
2
Figure 3. Backscattering from a body of revolution. The total scattered field consists of two components. One of them is a specular reflected ray (1) and the other is a beam of diffracted rays (2) radiated
by creeping waves traveling along the shadow side of the scattering
body. The equi-phase interference of these two components results in
the resonance-like increase of the scattered field.
177
continuity creates edge waves (B and C), which can be interpreted as diffracted rays. They represent the second-order
contributions (1215). Diffracted waves arising from corners
provide the third-order contributions (1214). At the shadow
boundary on a smooth scattering surface, the incident wave
excites creeping waves (D), which propagate along the shadow
side of the object and radiate surface diffracted rays (E). Due
to continuous radiation of these rays, the creeping waves attenuate exponentially and for this reason the contribution of
surface diffracted rays (E) to backscattering is small (12,13).
However, for the objects with dimensions comparable to the
wavelength, surface diffracted rays can give appreciable contributions, as it is mentioned already in the previous section
(see also Fig. 2). Diffracting waves propagating along the
scattering object can undergo multiple diffractions at geometrical and material discontinuities and can transform into
other types of waves. This process creates high-order contributions to backscattering (1215). A visual description of
scattering from large objects, simple quantitative estimations
of some contributions to backscattering, and relevant references are presented in (16).
High-frequency asymptotic methods are widely used to
predict scatterings in this frequency region. They include geometrical optics (GO) and its extension, geometrical theory of
diffraction (GTD); physical optics (PO) and its extension,
physical theory of diffraction (PTD); and various modifications and extensions of GTD and PTD. These asymptotic techniques are discussed in ELECTROMAGNETIC WAVE SCATTERING
and RADAR CROSS-SECTIONS. The present article supplements
these and concentrates mainly on the physical optics. This
method is not so precise as GTD, PTD, and their extensions,
but it allows useful estimations for the scattered fields in
which many practical problems cannot be treated with other
techniques. Direct numerical methods, in their classical
forms, are not efficient in the high-frequency region. Various
combinations of these methods with the asymptotic techniques (so-called hybrid methods) represent a promising direction in the prediction of high-frequency scattering (17). Additional information about numerical, hybrid, and asymptotic
techniques used for the solution of scattering problems can be
found in Refs. 6, 13, and 18 and in the reading list at the end
of this article.
B
A
D
B
C
E
E
178
yy
;;
BACKSCATTER
flected field. The reflection coefficient is found from the solution of an appropriate canonical problem. For opaque homogeneous objects, this is the Fresnel reflection coefficient, which
determines the amplitude and phase of plane waves reflected
from a planar boundary of a semi-infinite homogeneous medium [pp. 474479 of (2)]. For opaque objects coated with thin
layers, the canonical problem is the reflection of plane waves
from an infinite planar layer. This canonical layer is tangential to the scattering object (Fig. 6). It is homogeneous in the
directions parallel to its surface and has the same material
structure in depth as a real layer at the reflection point T.
The canonical layer is placed on the planar boundary of a
homogeneous medium with the same material properties as a
real object at the tangency point. This implies that the field
on a real coated object is determined exclusively by its local
properties in the vicinity of the reflection point. Nonlocal contributions from various waves propagating along the object
are not treated with this approach. Creeping and traveling
waves [pp. 120 and 130 of (2)] are examples of such waves.
According to this GO approach, the backscattering RCS of
smooth coated objects equals
= |r(0)| R1 R2
2
[(m) ]
(10)
(V/m)
je = n H
(11)
(A/m)
(12)
n E GO
j PO
m = n
This equation defines equivalent currents only on the illuminated side of the opaque scattering object. On the shadow
side, these currents are assumed to be zero. In the particular
case of perfectly conducting objects, the magnetic current does
not exist (jPO
m 0) due to the boundary condition n E 0,
and the electric current equals jPO
2n Hinc according to
e
the GO approximation.
The scattered field is found by the integration of equivalent
currents over the scattering surface S. Geometry of a sample
scattering problem is shown in Fig. 7, where the quantity r is
the distance between the integration (, , ) and observation
n
z
Canonical
Layer
Opaque object
with a layer
Figure 6. A scattered field at the reflection point on a coated scattering object is equal asymptotically (with kR1,2 ) to the field which
would be reflected from a tangential layer with the same material
properties. Due to losses, the contributions of rays and waves propagating along the object (inside the coating) become small and can be
neglected.
y
x
BACKSCATTER
179
2
(R, , ) points. In the far zone (R kmax
), the scattered field
is determined as
E = Z0 H = ik(Z0 Ae + Am
)
E = Z0 H = ik(Z0 Ae Am
)
ikR
1 e
A e,m =
j e,m eik cos dS
4 R S
cos = cos cos + sin sin cos( )
(13)
(15)
Accuracy of PO. Approximate estimations for the PO scattered field [Eq. (13)] can be found by the application of asymptotic techniques to the integrals [Eq. (14)] with the PO currents [Eq. (12)]. The first term of the asymptotic expansion
found in this way is correct for the fields scattered by smooth
convex objects and planar plates in the specular directions
predicted by GO. All higher-order terms in the PO asymptotic
expansion are incorrect. Only two exceptions exist when PO
provides the exact solution. The first is the scattering from
the infinite perfectly conducting plane. The second is the scattering from the semi-infinite perfectly conducting paraboloid
of revolution illuminated by the plane wave incident along
the symmetry axis (22). In the latter case, the scattered field
consists of only the GO reflected rays. Reference 22 also
shows that PO provides the correct second term in the highfrequency asymptotic expansion for the specular backscattering from any convex perfectly conducting bodies of revolution
when the incident wave propagates in the direction parallel
to the symmetry axis.
The first term of the PO asymptotic expansion for the field
scattered by smooth convex objects in specular directions represents the GO reflected rays [pp. 5062 of (2)]. Therefore, for
such objects the PO value of RCS in specular directions is
asymptotically (with k ) equivalent to the GO estimation.
However, it is well known that GO is valid only away from
the forward direction, i.e., from the shadow boundary of the
incident rays. But PO is more general than GO and is applicable in the vicinity of this direction. All known results show
that the first term of the PO asymptotic expansion for the
field scattered in the forward direction is correct and leads to
the following RCS for large opaque objects:
A2
2
[(m)2 ]
(14)
= 4
(16)
Here, the quantity A is the area of the scattering object projection on the plane perpendicular to the direction of the incident wave propagation.
A2
|re ( )|2
2
A2
h ( ) = 4 2 |rh ( )|2
e ( ) = 4
(17)
A2
|1 te ( )|2
2
A2
h ( ) = 4 2 |1 th ( )|2
e ( ) = 4
(18)
180
BACKSCATTER
1
2
Polarization of the PO Scattered Field. In general cases, components E and E in Eq. (13) have different amplitudes and
phases. This results in the elliptic polarization of the scattered field even in the case when the incident wave is linearly
polarized. This means that the electric vector of the scattered
field rotates with the angular frequency k c and its endpoint traces an ellipse. The direction of rotation (clockwise or
counterclockwise) is determined by the phase shift between
components E and E. The lengths of the elliptical axes are
determined by the amplitudes of these components. Due to
diffraction, the scattered field can contain an electric field
component perpendicular to the incident wave polarization.
This phenomenon is known as depolarization, or crosspolarization.
The PO field scattered by arbitrary perfectly conducting
objects in the backscattering direction does not contain the
crosspolarized component [p. 56 of (2)]. It is assumed only
that no multiple GO reflections occur on the objects surface.
This PO result is correct for scattering objects with certain
symmetry. These are objects with a symmetry plane parallel
both to the electric (or magnetic) vector of the incident wave
and to the direction of its propagation. Each element of such
a scattering object may create the crosspolarized component.
But due to the symmetry, the crosspolarized components from
symmetrical elements cancel each other in the backscattering
direction (Fig. 11). A convex smooth body of revolution whose
symmetry axis is parallel to the incident wave direction is a
simple example of such an object. A symmetrical plate, illuminated by the plane wave whose electric (or magnetic) vector
and direction of propagation are parallel to the symmetry
plane, is another example where the backscattered field does
not contain a crosspolarized component (Fig. 12).
As previously stated, the first term of the PO high-frequency asymptotic expansion represents the GO reflected ray.
This ray contains the crosspolarized component when the
electric vector of the incident ray makes any angle different
from 0 and 90 with the incidence plane at the reflection
point. This is the case when PO correctly describes depolarization of the scattered field in the high-frequency asymptotic
limit (k ). However, PO fails to predict all depolarization
effects caused by the diffraction part of equivalent surface
currents.
Ecr
1
Ecr
Einc
S
y
3
x
Figure 10. A scattered field is generated by the induced surface currents. Any approximations for these currents can result in the appearance of nonphysical components in the scattered field. In particular, the PO currents create spurious shooting-through edge waves (1,
2, and 3) passing through an opaque object.
Figure 11. Backscattering without depolarization from a symmetrical perfectly conducting surface S. The incident wave direction is parallel to the symmetry plane y-z. Vectors Ecr are the cross-polarized
components of the reflected field. Due to the symmetry, they cancel
each other.
BACKSCATTER
Einc
x
Figure 12. Backscattering without depolarization from a perfectly
conducting plate S. The incident wave direction is parallel to the symmetry plane y-z. Cross-polarized components scattered by the left and
right parts of the plate are symmetrical and completely cancel each
other.
Bistatic RCS. Bistatic RCS determines the power flux density of electromagnetic waves scattered by the object in an
arbitrary direction. The angle between the directions to the
transmitter and receiver is called the bistatic angle. Monostatic, or backscattering, RCS is a particular case of bistatic
RCS when the bistatic angle equals zero. Some PO results for
bistatic RCS have already been discussed. This section addresses the interrelationships between bistatic and monostatic RCS.
Using PO, one can prove the following statement:
For perfectly conducting bodies which are sufficiently smooth, in
the limit of vanishing wavelength, the bistatic cross section is
equal to the monostatic cross section at the bisector of the bistatic
angle between the direction to the transmitter and receiver.
[pp. 157160 of (1) and p. 11 of (2)]
(19)
It should be noted that this equation is valid also for the bistatic RCS, which therefore does not depend on the bistatic
angle 2 (Figs. 5 and 6). This follows directly from Eqs.
(5.32), (6.19), and (6.20), given in Chapter 8 of (19):
1
E (0) R1 R2
2
1
H (r) = H (0) R1 R2
2
E (r) =
eiks
s
eiks
s
181
(21)
where the reflection coefficients re,h() depend on the polarization and direction of the incident wave. Therefore, the asymptotic equivalence between bistatic and monostatic RCS can be
valid only for those bistatic angles where re,h() re,h(0).
This discussion relates to the situation where the bistatic
scattered field is produced by a single scattering source (the
reflection point). The field scattered by complex objects can
have many sources (scattering centers) on the objects surface.
Figure 13 illustrates the bistatic scattering from the center
located at the point (xn, yn, zn). The origin of Cartesian coordinates is somewhere inside the scattering object. The z-axis is
directed along the bisector of the bistatic angle . In the far
zone, the bistatic scattered field can be represented as the
sum of partial contributions from all scattering centers [pp.
983988 of (4)]:
u ( ) =
eikR
v n e2ikz n cos 2
R n
(22)
(xn , yn , zn) y
(20)
These expressions describe the field reflected by smooth convex objects at a far distance (s R1,2) from the reflection point
for any incidence angle (0 /2). In the case of reflection
from concave surfaces, the reflected field acquires the additional phase shift of (/2) in passing through a focus of reflected rays. Vectors E(0) and H(0) denote the reflected field
at the reflection point. Expressions (20) clearly show that the
/2
z
/2
R
Figure 13. Geometry of the bistatic scattering problem. A solid winding line represents a scattering object with many scattering centers
(xn, yn, zn). The scattering direction forms the angle with the direction of the incident wave.
182
BACKSCATTER
eikR
v n e2ikz n
R n
(23)
(24)
(25)
2a
l
Figure 14. Backscattering from a truncated cone. The base diameter
of the cone (2a) is large compared to the wavelength. The length of
the cone (l) can be arbitrary. In the limiting case l 0, the cone
transforms into a disk.
BACKSCATTER
(30)
2a
l
Figure 15. Backscattering from a truncated paraboloid. The base diameter of the paraboloid (2a) is large compared to the wavelength.
The length of the paraboloid (l) can be arbitrary. In the limiting case
l 0, the paraboloid transforms into a disk.
2
1
tan2 sin kl tan eikl
PO = a2
ka
(26)
where the cone length equals l a cot . To clarify the physics in this equation, we rewrite it as
2
i
i
tan eikl tan +
tan eikl
PO = a2
2ka
2ka
(27)
2
2
1
sin
2
PTD
2
n
n
ikl
tan sin kl +
e
= a
2
ka
cos cos
n
n
(28)
(31)
2
2
sin
PTD
2
n
n
2ikl
e
= a tan +
2
cos cos
n
n
(32)
tan
PO = a2
e2ikl
cos
2ka
2ka
(33)
PO
2
1
2ikl
tan e
= a
cos
2
183
(34)
(29)
2a
l
Figure 16. Backscattering from a truncated sphere. The base diameter of the sphere (2a) is large compared to the wavelength. The length
of the sphere (l) can be arbitrary. In the limiting case l 0, the
sphere transforms into a disk.
184
BACKSCATTER
2
2
1
sin
n
n
+
e2ikl
PTD = a2
2
cos
cos cos
n
n
(35)
BACKSCATTER
185
BIBLIOGRAPHY
1. J. W. Crispin Jr. and K. M. Siegel (eds.), Methods of Radar CrossSection Analysis. New York: Academic Press, 1968.
2. G. T. Ruck, D. E. Barrick, W. D. Stuart, and C. K. Kirchbaum,
Radar Cross-Section Handbook, Vol. 1 and 2. New York: Plenum
Press, 1970.
3. J. J. Bowman, T. B. A. Senior, and P. L. E. Uslenghi (eds.), Electromagnetic and Acoustic Scattering by Simple Shapes. New York:
Hemisphere Publishing Corp., 1987
4. P. C. Fritch (ed.), Special Issue on Radar Reflectivity, Proc. IEEE,
53 (8): August 1965.
5. M. N. O. Sadiku, Numerical Techniques in Electromagnetics. Boca
Raton, FL: CRC Press, 1992.
6. E. K. Miller, L. Medgyesi-Mitschang, and E. H. Newman, Computational Electromagnetics, New York: IEEE Press, 1991.
7. P. P. Silvester and G. Pelosi (eds.), Finite Elements for Wave Electromagnetics, New York: IEEE Press, 1994.
8. P. Ya. Ufimtsev and A. P. Krasnozhen, Scattering from a straight
thin wire resonator, Electromagnetics, 12 (2): 133146, 1992.
9. L. A. Vainshtein, Waves of current in a thin cylindrical conductor, II. The current in a passive oscillator, and the radiation of a
transmitting antenna, Sov. Phys., Tech. Phys., 4 (6): 617626,
December 1959.
10. L. A. Vainshtein, Waves of current in a thin cylindrical conductor, III. Variational method and its application to the theory of
ideal and impedance conductors, Sov. Phys., Tech. Phys., 6 (1):
1929, July 1961.
11. M. A. Leontovich and M. L. Levin, Towards a theory on the simulation of oscillations in dipole antennas, Zhurnal Technicheskoi
Fiziki, 14 (9): 481506, 1944 (in Russian). [The English translation is published in the report: K. C. Chen (ed.), SAND91-0720,
UC-705, Sandia National Laboratories, Albuquerque, NM and
Livermore, CA, Contract DE-AC04-76DP00789, January 1992].
12. J. B. Keller, Geometrical theory of diffraction, J. Opt. Soc. Am.,
52: 116130, 1962.
13. R. C. Hansen (ed.), Geometrical Theory of Diffraction, New York:
IEEE Press, 1981.
14. P. Ya. Ufimtsev, Method of Edge Waves in the Physical Theory of
Diffraction. Moscow: Soviet Radio Publishing House, pp. 1243,
1962. Translated by U.S. Air Force, Foreign Technology Division,
Wright-Patterson AFB, Ohio, 1971; Technical Report AD No.
733203, DTIC, Cameron Station, Alexandria, VA.
15. P. Ya. Ufimtsev, Elementary edge waves and the physical theory
of diffraction, Electromagnetics, 11 (2): 125160, 1991.
16. P. Ya. Ufimtsev, Comments on diffraction principles and limitations of RCS reduction techniques, Proc. IEEE, 84 (12): 1830
1851, 1996.
17. D. P. Bouche, F. A. Molinet, and R. Mittra, Asymptotic and hybrid techniques for electromagnetic scattering, Proc. IEEE, 81:
16581684, 1993.
18. W. R. Stone (ed.), Radar Cross Section of Complex Objects, New
York: IEEE Press, 1990.
19. V. A. Fock, Electromagnetic Diffraction and Propagation Problems. London: Pergamon Press, 1965.
20. M. Born and E. Wolf, Principles of Optics, New York: Pergamon,
1975.
21. H. M. Macdonald, The effect produced by an obstacle on a train
of electric waves, Phil. Trans. Roy. Soc. Lond., Series A, Math.
Phys. Sci., 212: 299337, 1912.
22. C. E. Schensted, Electromagnetic and acoustic scattering by a
semi-infinite body of revolution, J. Appl. Phys., 26: 306308,
1955.
186
BACKSCATTER
23. P. Ya. Ufimtsev, Diffraction of electromagnetic waves at blackbodies and semi-transparent plates, Radiophys. Quantum Electr.,
11: 527538, 1968.
24. P. Ya. Ufimtsev, Comments on Comparison of three high frequency diffraction techniques, Proc. IEEE, 63: 17341737, 1975.
25. R. D. Graglia et al., Electromagnetic scattering for oblique incidence on impedance bodies of revolution, IEEE Trans. Antennas
Propag., 43 (1): 1126, 1995.
26. M. W. Browne, Two Rival Designers Led the Way to Stealthy
Warplanes, in The New York Times, Science Times Section,
US, May 14, 1991.
27. S. F. Brown, The Secret Ship, in magazine Popular Science,
US, October 1993.
28. B. Rich and L. Janos, Skunk Works, Boston-New York-London:
Little, Brown & Company, 1994.
29. K. J. Vinoy and R. M. Jha, Radar Absorbing Materials, Boston:
Kluwer Academic Publishers, 1996.
Reading List
This section contains short comments on some related references.
P. C. Fritch (ed.), Radar Reflectivity, Special issue of the Proc. IEEE,
53 (8), August 1965.
The first attempt to sum up basic results in the field of RCS.
Includes a comprehensive subject index, about 1500 titles (pp.
10251064).
J. W. Crispin Jr. and K. M. Siegel (eds.), Methods of Radar Cross
Section Analysis. New York: Academic Press, 1968.
Includes a short historical survey of high-frequency approximations. Contains results of such approximations for monostatic and
bistatic RCS of simple objects. States a strategy of RCS calculations for complex objects. Some results are conveniently summarized in tables. Table 5 on p. 147 contains expressions for RCS in
the Rayleigh region (for objects small in comparison with wavelength). High-frequency monostatic and bistatic RCS are given in
Tables 7 and 8 on p. 168, 169, 171.
G. T. Ruck, D. E. Barrick, W. D. Stuart, and C. K. Kirchbaum, Radar
Cross-Section Handbook. New York: Plenum Press, 1970.
This is a real encyclopedia of RCS, which includes most results
obtained before 1970. It contains numerous theoretical and experimental results for both perfectly conducting and absorbing objects.
The physical theory of diffraction (PTD) is mentioned here as the
SommerfeldMacdonald technique. Equation (5.1-54), presented
on p. 351, for the bistatic RCS of an ellipsoid is incorrect. In the
particular case when an ellipsoid transforms into a sphere, this
equation does not provide the bistatic RCS for the sphere,
a2. Instead it leads to the wrong quantity a2 /(1 cos )2,
where is the bistatic angle.
J. J. Bowman, T. B. A. Senior, and P. L. E. Uslenghi (eds.), Electromagnetic and Acoustic Scattering by Simple Shapes. New York:
Hemisphere Publishing Corp., 1987.
Contains a comprehensive collection of theoretical results for
RCS of simple objects which allow the exact solutions of diffraction
problems. Both low-frequency and high-frequency approximations
are presented as well.
M. Skolnik (ed.), Radar Handbook. New York: McGraw-Hill, 1970.
Contains many results of measurements and calculations for
RCS. Calculations were carried out mostly by the physical optics
approach and GTD. Analytical expressions for RCS are not given.
W. R. Stone (ed.), Radar Cross Sections of Complex Objects, Special
issue of the Proc. IEEE, 77 (5), May 1989.
W. R. Stone (ed.), Radar Cross Sections of Complex Objects, Special
issue of the IEEE Trans. Antennas Propag. 37 (5), May 1989.
These two references contain many theoretical results concerning RCS for complex objects (perfectly conducting objects with
BACKTRACKING
This book consists of a collection of articles. It contains key
papers on GTD, asymptotic solutions of some canonical problems,
and applications-oriented papers.
Many scattering objects contain nonmetallic materials, composites, and various layered structures. To simplify the solution of
scattering problems for such objects, it is often practical to apply
approximate boundary conditions. These conditions are enforced
on the external surface of the object and contain important information about the internal structure of the scattering object. As a
result, this approximation allows one to substantially reduce the
spatial region under investigation. The two following books present the development and applications of this approximation technique.
T. B. A. Senior and J. L. Volakis, Approximate Boundary Conditions
in Electromagnetics. London: The Institution of Electrical Engineering, 1995.
D. J. Hoppe and Y. Rahmat-Samii, Impedance Boundary Conditions
in Electromagnetics, Washington, D.C.: Taylor & Francis, 1995.
187
CABLE TELEVISION
Cable TV (CATV) as a successful business can be considered an American invention. Even though first commercial cable television installations date back as early as 1936 in Europe (United Kingdom and Germany),
they did not create a market at that time.
Right after World War II, the first CATV systems in the United States appeared almost simultaneously
in Pennsylvania and Oregon. Community antenna television was conceived as a system for delivery by cable
of signals broadcast by wireless television stations. The system was primarily targeted to deliver television
in rural areas with poor wireless reception. A community antenna provided broadcast television to these
neighborhoods via coaxial cables. The abbreviation CATV meant community antenna television at this early
stage of the system.
For distant places the operator used several microwave links to bring the three major network signals
across hundreds of kilometers and then distributed them around the local community by a cable network. If
mountains blocked the radio path, a single antenna on the mountaintop fed the cable trunk down into the town.
This kind of service was known as basic cable, and the customer paid few dollars per month for a high-quality
TV signal of three existing channels. Advertisers on those channels were the real promoters of CATV at that
time.
The need for signal amplification approximately every kilometer limited the area that could be served.
Also, a separate amplifier was needed for each channel, which restricted the number of channels to three or
four. In the mid-1960s, improvements in transistors boosted channel numbers progressively. By 1970 there
were 2750 operators serving six million homes.
In 1974 satellites (so far used only for government purposes) became commercial. Cable operators started
to use satellite transponders to rebroadcast to their facilities programs from multiple producers, and their large
capacity made it possible to offer many channels using up to a total of 300 MHz of bandwidth. Service providers
started to produce programming and sell their services to cable as well as to direct broadcast satellite (DBS)
operators, thus motivating rapid deployment in large metropolitan areas, where cable distribution systems
competed with DBS.
Networks specifically designed to be distributed by the cable system appeared in the United States by
1975 with Time Inc.s Home Box Office (HBO); soon others emerged, such as Ted Turners Superstation (soon
renamed WTBS) in 1976, and C-SPAN (live broadcasts of the House of Representatives), ESPN (sports), and
Nickelodeon (childrens programming), all in 1979. The 24-hour news Cable News Network (CNN) followed in
1980.
Though CATV offered a wider range of programming than ordinary broadcasting, the capacity of pure
coaxial systems became insufficient, being still small compared to the number of DBS channels. To overcome
this lack, with the invention of the linear light source in the 1980s, cable operators started to replace coaxial
cable trunks with fiber systems. The benefit was not only in reducing the number of amplifiers (thus increasing
available bandwidth) but also in improving the reliability while minimizing maintenance costs. These networks
are referred to as CATV HFC networks, where CATV no longer means community antenna television but cable
television, and HFC stands for hybrid fiber coaxial. New channels were pay-per-view, bringing in new revenues
1
CABLE TELEVISION
to broadcasters while broadening opportunities to the creative community. This development reached a peak
by the mid-1980s and congestion by the early 1990s.
It is worth noting that cable operators must negotiate franchises with municipalities that authorize them
to lay cable in the ground, the use of poles, and so on, in exchange for fees to the local government. In this
respect CATV is similar to public utilities, since it uses public rights and places to deploy a capital-intensive
network. It distributes services from content providers, and subscribers are billed on a monthly basis.
CATV networks have their main market in United States. Their development followed different patterns
in other countries. CATV penetration in Southern European countries is still small, but current deployments
are full digital and with a large optical trunk plant to target directly not only broadcasting but the broadband
market. CATV is full-grown in Central European countries such as in Germany.
Worldwide, CATV networks serve over 200 million residences, while telephony serves more than 500 million. CATV is currently extending its capabilities by migrating from analog narrowband to digital broadband.
Indeed, it has recently emerged as a promising access network infrastructure for the delivery of voice, video,
and new broadband applications. Current research efforts are focused on the design of protocols for CATV to
deliver different levels of quality of service (QoS) for diverse user applications.
Migration to digital improved the system capacity substantially, since every 6 MHz analog video channel
yields 27 Mbit/s of raw throughput with 64-QAM modulation and RS FEC. Such a data rate can convey up to six
digital video channels through MPEG-2 compression with the same quality as analog channels. Migration to
CABLE TELEVISION
Fig. 1. CATV architectures: (a) tree-and-branch all-coaxial network; (b) hybrid fiber coaxial network with an optical trunk
ring.
hybrid coaxial has increased robustness by reducing the number of households in serving areas and increasing
the bandwidth (>750 MHz) by decreasing the coaxial cascade depth (1).
Figure 1(a) shows an early tree-and-branch all-coaxial CATV network architecture. Figure 1(b) shows
the upgrading to an optical trunk ring (HFC). The dowstream traffic flows from the headend and is directly
injected into the trunk fiber (originally coaxial) ring. The optical section is terminated with a Hub, or node,
which converts optical signals to electrical. Feeder coaxial cable distribution is arranged from the node in a
tree-and-branch topology where traffic is split at branching points to up to 2000 houses. Propagation through
the cable attenuates the signal by an amount proportional to the square root of frequency, resulting in higher
losses for higher frequencies, which is mitigated with amplifiers and equalizers, especially in the UHF bands.
The HFC access network represents the natural evolution of the existing mature tree-and-branch CATV
networks for introducing the new digital broadcast or interactive services. An HFC network provides transparent communication channels between the subscribers and the CATV hubs or headends, since no signal
processing occurs between the subscriber and the hub, besides optical-to-electrical conversion and eventual
frequency translation and amplification. The hub provides the interface between the core network and the
HFC distribution network. At this level, communication equipment is used to transmit an analog frequency
multiplex comprising both the TV broadcast signals (analog or digital) and the modulated digital interactive
ones. Each node generally supplies up to 1000 subscribers (2,3).
CABLE TELEVISION
Figure 2 shows a scheme of the headend, comprising a number of possible elements and functionalities
of a two-way digital CATV HFC network. Analog or digital broadcast TV is not only the content source of the
cable system, which can also supply video and data with local or remote (Internet) servers. Telephone services
are also contemplated through connection to the PSTN.
Figure 3 shows the elements at user premises. Different modems are necessary for video and data
applications, and also an MPEG-2-compliant set-top decoder is needed to transform the digital transport
stream for the analog TV set to be able to present the image.
CABLE TELEVISION
Fig. 3. CATV elements for data and video applications at user premises.
Each channel carries a potential payload of 30 to 40 Mbit/s using a complex multilevel modulation scheme
(64-QAM typical).
Assuming downstream channels of 6 MHz with an effective bandwidth of 5.4 MHz and channels from
550 MHz to 750 MHz, the total capacity is approximately of 1 Gbit/s, which should be sufficient at least until
high-definition TV (HDTV) arrives. For the upstream capacity let us assume also 6 MHz channels with QPSK
modulation, as is being proposed by standardization groups, and a total bandwidth of 30 MHz (more or less
as shown in Figure 4). With these assumptions the upstream capacity becomes 10 Mbit/s per channel and 60
Mbit/s in total. Taking into account up to 20% of overhead for upper-layer protocols, the upstream capacity
may be reduced to 5% of downstream capacity. This worst-case capacity may be enough, though, since peak
CABLE TELEVISION
hours can be assumed to be below 25% of the day. However, voice applications, for example, need real-time
delivery, taking priority over available TCP/IP connections. Bandwidth assignment and bit-rate management
by differentiating best-effort traffic from traffic with guaranteed bandwidth are QoS problems to be solved by
cable networks. Interested readers may refer to Refs. 4 and 5.
Upstream Transmission
CATV networks were designed for video broadcasting, but cable operators soon realized the usefulness of
a return path for movie selection, pay-per-view, billing information, monitoring, and so on. One-way cable
networks had an inherent capability for upstream transmission, since raw wire can transport signals in both
directions simultaneously. However, amplifiers boost signals only in one direction, blocking the return path.
Thus, upgrading to the two-way network architecture must include two-way amplifiers, laser transmitters to
the headend, and additional equipment at the user premises. It should be noted that downstream video may
be broadcast or on demand, but the reverse path is a shared medium, which means access must be controlled
via some medium access (MAC) protocol (6).
An important issue for the reverse path is ingress noise degrading the quality of service. Home electrical
devices (hair dryers, vacuum cleaners, etc.) create noise bursts within the 5 MHz to 40 MHz frequency range
(within the upstream bandwidth; see Fig. 4), which are propagated back to the network through the amplifiers.
Amateur radio and AM radio are also potential sources of interference in the return path. A number of
techniques are used to mitigate the interference and noise present in the upstream spectrum:
Low-pass filtering at the cable drop in the data terminal in the home
Modulation techniques such as frequency hopping and spread-spectrum cable modems
Standards
The rapid evolution of broadband service has resulted in cable television operators, providers of telephone
service, and other service providers seeking to provide voice, data, and video services to their residential
and commercial subscribers over existing and new infrastructures. In this context, standardization is needed,
mainly to make possible retail distribution of cable network elements in the home for both video and data
applications. The standardization of technological features of these elements provides a means to reduce costs
by creating a competitive market. Standardization activities cover not only the physical layer (PHY), but also
the MAC layer, security, and service management. Organizations involved in cable TV standardization are the
following (1):
CABLE TELEVISION
Davic And Dvb. DAVIC was started by Leonardo Charliglione, the Motion Picture Experts Group
(MPEG) committee head. It was established in 1994 as a nonprofit association based in Switzerland, and it was
constituted with companies involved in all sectors of the audiovisual industry (manufacturing and service),
as well as a number of government agencies and research organizations from more than 25 countries. The
association was disbanded after 5 years of activity, remaining active only through its Web site (www.davic.org).
DAVIC specifications were primarily aimed at promoting the success of interactive digital audiovisual applications and services, and therefore they are based on DVB specifications. Since DAVIC aims at interoperability
across applications, it does not specify systems but components (tools) that are non-system-specific in that they
have to be usable by different industries in different systems and still guarantee interoperability.
The DVB Project is a consortium of companies from more than 30 countries worldwide in the fields of
broadcasting, manufacturing, network operation, and regulatory matters. The DVB Project Office is based in
Geneva, Switzerland. It aims at establishing common international standards for the migration from analog to
digital broadcasting through the design of a global standard for the delivery of digital television (www.dvb.org).
Their standards are based on the common MPEG-2 coding. Through the use of MPEG-2 stream packets, DVB
is potentially able to deliver TV service [from HDTV to multiple-channel standard-definition TV (PAL/NTSC
or SECAM)], broadband multimedia data, and interactive services effortlessly from one medium to another. In
particular, DVB signals can move from satellite to cable and from cable to terrestrial. DVB standards are published by the European Telecommunications Standards Institute (ETSI). ETSI, the Centre for Electrotechnical
Standards (CENELEC), and the European Broadcasting Union (EBU) have formed a joint technical committee
(JTC) to handle the DVB family of standards, which can be grouped as follows with regard to transmission
standards:
Specifications 1.2 to 1.4 were issued subsequently. The last specification, 1.5, dealt with specific technical
issues:
CABLE TELEVISION
DVB has submitted to ETSI a complete set of standards defining issues related to the provision of
interactive services on cable and LMDS networks:
The DVB RCCL [return channels for cable and local multipoint distribution systems (LMDSs)] includes the
specifications related to network-dependent layers and has been developed in close consultation with DAVIC.
The result is the adoption by DAVIC of ETS 300 800 as the only standard for cable modem applications. This
is a major step towards common specifications between DVB and DAVIC.
Working in this direction, the DVB/DAVIC Interoperability Consortium provides multivendor interoperability based on the open international standards DVB RCCL (ETSI ETS 300 800) and DAVIC 1.5 for broadband
communication (www.dvb-davic.org). The Consortium intends to build a multiservice platform enabling a single system to deliver video, audio, data, and voice to residential devices such as cable modems, set-top boxes,
multimedia home platforms, and residential gateways. The consortium of European multisystem operators
(MSOs) has recognized it as the preferred technology, so that the DVB-RCCL/DAVIC specification represents
a technology that may become a standard for international deployment.
The following vendors have announced their intention to comply with the specification: Alcatel, Cisco,
DiviCom, Hughes Network Systems, Nokia Multimedia Network Terminals, Sagem, The Industree, Thomson
Broadcast Systems (a subsidiary of Thomson Multimedia), and Thomson Multimedia. It is apparent that while
European MSOs have committed to DVB-RCCL/DAVIC, other big providers have already chosen to deploy
DOCSIS (data-over-cable service interface specification) technology. DVB-RCC is available as ETSI standard
ETS 300 800, Digital Video Broadcasting (DVB); Interaction Channel for Cable TV Distribution Systems
(CATV). DVB-C is available as ETSI standard ETS 300 429, Digital Video Broadcasting (DVB); Framing
Structure, Channel Coding and Modulation for Cable Systems.
Key physical-layer features of specifications for HFC networks are the following:
Upstream:
CABLE TELEVISION
Fig. 5. DOCSIS architecture model. CM: cable modem; CMTS: cable modem termination system.
Docsis And Mcns. MCNS is a consortium of North American entities that came together to develop
the specification DOCSIS for transmitting data over a cable network (www.cablemodem.com). DOCSIS 1.0 was
proposed in March 1997. In March 1998 the ITU accepted DOCSIS as a cable modem standard (J.112). To
deliver DOCSIS over a cable television (CAT) network, one 6 MHz RF channel in the 50 MHz to 750 MHz
range is typically allocated for downstream traffic to homes, and another channel in the 5 MHz to 42 MHz
band is used to carry upstream signals. Cable modems were described as external devices that connect to a
personal computer through a standard 10Base-T (10 Mbit/s Ethernet over CAT3-5 cabling) Ethernet card or
USB interface. CableLabs manages a certification process to ensure DOCSIS cable modems manufactured by
different vendors comply with the standard and are interoperable.
CableLabs is an organization dedicated to the certification of DOCSIS-based equipment. In April 1999 it
issued a second-generation specification called DOCSIS 1.1, which adds key enhancements to the original standard, such as improved QoS and hardware-based packet-fragmentation capabilities, to support IP telephony
and other constant-bit-rate services. The next-generation standard is designed to be backward compatible, enabling DOCSIS 1.0 and 1.1 modems to operate in the same spectrum on the same network. A third-generation
DOCSIS standard will add an advanced PHY to the core specification to increase the upstream transmission
capacity and reliability by using FA-TDMA (frequency-agile TDMA) and S-CDMA (synchronous code division
multiple access).
Vendors supporting the DOCSIS standard are 3Com, Bay Networks, Cisco Systems, General Instruments,
Hewlett-Packard, Hybrid Networks, Intel, Motorola, NEC, Panasonic, Scientific Atlanta, Sharp Electronics,
Toshiba, US Robotics, and Zenith Electronics.
The DOCSIS architecture model is shown in Figure 5. The PC generates Internet protocol (IP) over
Ethernet packets. The cable modem (CM) acts as a bridge and forwards the Ethernet frames to the network.
DOCSIS specifies a new MAC layer from the CM to the cable modem termination system (CMTS) in the
upstream direction. The Ethernet frame is encapsulated by the CM in a DOCSIS MAC frame and sent to the
CMTS. The CMTS (a router or a bridge) de-encapsulates the Ethernet frame and forwards it upstream.
10
CABLE TELEVISION
The relationship between a CM and its CMTS is a master-slave relationship. The CMTS controls the
bandwidth allocation on the upstream channel. The CMTS sends on the downstream channel bandwidth
allocation messages called upstream bandwidth allocation maps (referred to as MAPs), which define how the
time units (mini-slots) on the upstream channel must be used. The cable modem boot process used by the CM
influences architecture design, and DOCSIS concludes with an explanation of how the CM boots up.
Key PHY features of specifications for HFC networks are the following:
Upstream:
Downstream:
DOCSIS standards have not yet been accepted in Europe and Asia. The alternative standard DAVIC/DVB
for cable modem systems, also called the EuroModem standard, has been formalized, and, as was stated before,
a significant number of European cable system operators have embraced it.
IEEE 802.14. The IEEE 802.14 Working Group is a committee of engineers representing the vendor
community that has developed a specification for data-over-cable networking. The group was formed in 1994
and intended to develop a specification that would be recognized as an international standard. However, MCNS
defined an international specification first. Despite ITU recognition of DOCSIS as an international standard in
1999, the IEEE 802.14 Working Group continued to work on its specification, and MCNS indicated that it would
implement IEEE 802.14s advanced PHY specification. Overall the future of the IEEE 802.14 specification
is unknown. The group had good intentions and its specification was undoubtedly a better technological
alternative than that developed by MCNS. However, timing is critical in developing standards.
At the current time, performance characteristics of the standardized DOCSIS and the DVB/DAVIC CM
specifications have not been well assessed. For both specifications an OPNET model has been created by
CableLabs and MIL3 for the former, and by EuroCableLabs Centre of Competence for the latter. However, both
models support only the features of the upstream channel. Current research aims at the development of a new
simulation model for the downstream channels for both the DOCSIS and the DVB/DAVIC protocols using the
OPNET simulation package. Its performance should be based on the system throughput versus offered load
and the end-to-end delay versus system throughput with respect to:
(1) Number of CMs in the CATV network
(2) Variable offered load
(3) Different scheduling algorithms.
CABLE TELEVISION
11
been different, though it is expected that these two courses will probably converge in some way. Table 1 shows
the technical differences between the standards. Recently OpenCable has started to design a retail-model
set-top box using the DOCSIS standard for data, but also using the DAVIC standard. This box has DAVIC as
well as DOCSIS components: DAVIC is chosen as a core requirement, and DOCSIS as an extension for the
specification.
Applications
Two-way CATV HFC applications should be grouped into two major markets, in part because such applications are driving different (not necessarily divergent in principle) paths in the technology upgrading and
standardization:
Digital video over cable TV networks (although in the transition from analog to digital broadcasting, cable
systems will also deliver analog channels)
Data services over cable TV networks.
In fact the two markets will converge, since multimedia applications relate to both. The road to such
a confluence is being paved now through the standardization process, and the success achieved by each of
the parts will depend on both cost and technological accomplishments. In either case digital video or data
application services are to be delivered both in broadcast and in interactive mode.
12
CABLE TELEVISION
Digital Video over Cable TV Networks. First it is interesting to point out differences and key
technological and performance aspects of analog versus digital video. Unlike analog video, which degrades in
quality, digital video is virtually identical to the parent. There is a major difference between the way computers
and television handle video. When a television set displays its analog video signal, it displays the odd lines (the
odd field) first and the even lines (the even field) next. Each pair forms a frame, and there are 60 of these fields
displayed every second (30 frames every second). This is referred to as interlaced video. Instead, the computer
displays each line in sequence, from top to bottom. This entire frame is displayed 30 times every second. This
is often referred to as noninterlaced video.
Analog formats include NTSC (National Technical Standards Committee), used in the United States,
Mexico, Japan and Canada. Phase alternation line (PAL) is used in western Europe, and Sequential Couleur a`
Memoire (SECAM) is used in France, Russia, and Eastern Europe (Hewlett Packard, p. 4).
Digital formats include Joint Photographic Experts Group (JPEG), Motion JPEG (MJPEG), MPEG, Advanced Video Interleave (AVI) from Microsoft, Indeo from Intel, and CellB for use on Sun SPARCstations. Of
these digital formats, only MPEG is an internationally recognized compression standard.
Benefits of digital video are in random access to stored information, the ability to compress the video,
and the low cost of reproduction without loss of quality. From a transmission point of view, a number of digital
channels can be sent where only one analog channel was allocated. VCR video is in analog format running
at 30 frames/s. To achieve the same quality on the desktop, without compromising on frame rate, resolution,
color resolution, or image quality, one needs hundreds of gigabytes of disk space. Twenty-five centimeters (10
inches) of videotape is required to record one second of video. Compression is the solution, and there are two
types:
Intraframe compression compresses each individual video frame. JPEG and MJPEG use intraframe for
compression ratios of 20 : 1 to 40 : 1. The larger the ratio, the poorer the quality.
Interframe compression only looks at the video data that have changed. This also reduces the bandwidth
necessary to carry the video stream. MPEG uses interframe compression and can achieve up to 200 : 1
compression rates. Roughly, 9 Mbyte of hard-disk space is required for every minute of MPEG compressed
video. CATV and DBS use MPEG-2.
TV broadcasting and video conferencing, which allows a group of users to communicate in real time
through the use of live streaming audio and video, are the main applications of digital video.
Data Services over Cable TV Networks. Data services can be divided into narrowband and broadband applications according to the amount of bandwidth required. Telephony is narrowband, and multimedia
contents are broadband.
Telephony. We may include voice services within data interactive applications. Telephony can be offered
by cable by using IP to deliver voice through the use of a gateway located at the cable headend. Telephony
networks can either be private or allow access across the Internet or to and from the PSTN. Videotelephony is
a natural successor to telephony, although more complex. There is a tradeoff between quality, bandwidth, and
real-time.
PUSH Applications. PUSH technology is the delivery of information from server to client over IP.
Webcasting and multicasting (transmission of files or streaming audio and video to preselected multiple users)
are PUSH applications.
Virtual Private Networks. Since a cable system acts as a broadband network, operators are capable of
establishing virtual private networks (VPNs) as completely closed environments (intranets) allowing users to
access the Internet.
CABLE TELEVISION
13
Home Monitoring and Security. Cable networks can be used to monitor homes for fires, break-ins, and
medical emergencies. The available bandwidth of a cable network, combined with the systems being always
on, enables emergency information to be quickly transmitted.
Telemedicine. Medical doctors, patients, hospitals, clinics, and mobile units can use the CATV broadband
technology for quick transfer of huge files to provide diagnosis, treatment, consulting, and education.
Immersive Environments. An immersive environment is a virtual social environment, managed by a
computer program, where the user is represented within the program as an animated character (avatar)
visible to all of the users. This is clearly a broadband multimedia application, where sound, data, and video
come together, requiring a great amount of bandwidth along with critical constraints on latency.
One goal for a media immersion environment (MIE) is for people to interact, communicate, collaborate,
and entertain themselves naturally in a shared virtual space while they reside in distant physical locations.
The MIE has applications in many domains where physical presence is expensive (e.g., distance learning);
impossible (e.g., space exploration), unsafe (e.g., nuclear studies), or inconvenient (e.g., entertainment), or
where more than one person must be involved (e.g., remote medicine). The principal function of MIEs is to
synthesize multimodal perceptions that do not exist in the current physical environment, thus immersing users
in a seamless blend of visual, aural, and (potentially) haptic information.
Alternative Technologies
A CATV HFC is called an access network; it may also be referred to as a last-mile solution or (for solutions
proposed by telephone companies) a local loop solution. An access network connects customers premises to
the network termination and performs operations interfacing with the transport network, content provider,
and home network elements. Figure 6 shows graphically the technological frame for the access network. All
technologies share a common element, which is the network termination (NT), or network interface, in the
home. The main functions of these networks are the following:
A number of alternative cable and wireless competitive technologies are becoming feasible for the access
network, so that major CATV providers have had to accelerate standardization of their products.
Competitive wire technologies are the following:
xDSL (asymmetric, symmetric, high- and very high-bit-rate digital subscriber line)
FTTx (fiber to the neighborhood, curb, building, etc.)
ISDN (Integrated Service Digital Network).
Competitive wireless technologies are the following:
14
CABLE TELEVISION
xDSL (Digital Subscriber Line). xDSL refers to a series of networking technologies, comprising ADSL,
HSDL, VDSL, and SDSL, that are capable of supporting high data rates over the existing telephone network.
The technology bases its competitiveness on the existing extensive infrastructure and heavy capitalization
associated with telcos.
Standard telecom modems establish a data stream between two arbitrary points using the entire telecom
systemthat is, from the senders local loop, through the telephone switching system (mostly digital switches
now), and then to the receivers local loop. Standard modem connections can span continents, with one end
thousands of kilometers from the other end. DSL modems, on the other hand, establish a connection from
one end of a copper wire to the other end of that copper wire: the signal does not pass into the telephone
switching system. Consequently, DSL modems are not limited to using the voice frequencies passed by the
standard telephone system (typically 0 to 4 kHz); DSL modems typically use more than 100 kHz. To reiterate,
one end of the DSL link will be at the consumer site, the other end must be at the other end of the copper cable,
usually at the local telephone exchange, where data and voice are split. The voice frequencies are wired into a
traditional plain ordinary telephone service (POTS) switch and enter the usual telephone switching network.
The data frequencies are wired into a corresponding DSL modem, and the resulting high-speed digital data
stream coming from (or going to) the consumer is then handled as ordinary data (not analog voice) and may be
hooked into any number of networking technologies for further connection to the datas destination. Thus, the
data never enter the standard telephone switching system. Typically the data will be routed over a local-area
network (LAN) or wide-area network (WAN) connection (10Base-T Ethernet, T1, T3, ATM, frame relay) to a
business office.
Over the next five years, xDSL (and particularly ADSL) is considered to be the greatest threat to the cable
modem industry.
FTTx (Fiber to the Neighborhood, Curb, Building, Etc.). The growing demand for interactivity and
more bandwidth per subscriber is being satisfied by pushing fiber closer to the home and by the availability
of the required electro-optic components. FTTx refers to a series of networking technologies that run optical
CABLE TELEVISION
15
fiber from the central office to a users neighborhood (FTTN), curb (FTTC), building (FTTB), home (FTTH), etc.
Though the optical fiber is relatively inexpensive, optical transmitters and receivers are very costly. However,
continued growth of broadband services is fueled by the clear advantages that optical fiber systems offer in
cost, reliability, and performance for broadcast networks.
FTTH is not a practical solution for the moment to deliver residential data services. The present idea is
to replace long copper lines with fiber optic lines (not only in the telephone network, but also in CATV). The
major drawback is the cost of the replacement of the copper-based infrastructure. A critical point is what part
of the network will be replaced with fiber optic lines: that is what distinguishes FTTH, FTTB, fiber to the office,
FTTN, fiber-to-the-street, FTTC, etc. On the other hand, users of a CATV network again use copper lines, but
all the rest of the infrastructure is fiber-optics-based. With the exception of FTTH, all the above approaches
use some form of high-speed metallic access technology for service delivery to the customers premises.
ISDN (Integrated Service Digital Network). ISDN is a service provided by local telephone companies
that modifies regular telephone lines so that they can transmit data almost five times as fast as the fastest
analog modems currently available. In addition to the significant increase in transmission speed, ISDN also
allows the transmission of not only data, but a combination of data, voice, and video simultaneously on one
line. ISDN provides higher speeds than POTS by allowing data to be transferred digitally from end to end.
In contrast, POTS converts the digital data to analog within the local loop that extends into the home or
office, significantly reducing transmission speed. An ISDN line can carry up to 128 kbit/s of data. Converting
to an ISDN-compatible configuration for the consumer only requires an additional piece or two of relatively
inexpensive hardware. Telecommuting to an office or base requires additional equipment at the base. While
ISDN usage costs are slightly higher than those for analog telephone lines, users enjoy more than commensurate
benefits.
Though ISDN has a share of the high-speed connectivity market, the technology is limited to 128 kbit/s
and thus is in a different class than technologies that can support megabytes per second. In particular, it is
very limited for video applications; it can only be used for videoconferencing with 6 to 8 frames per second.
Satellites. There are two basic types of satellite systems being proposed: GEO and LEO.
GEOs orbit in the Clarke belt, approximately 35,000 km (22,000 miles) above the equator. With this
orbit, the satellite can stay over the same area of the earth for an indefinite period of time. Each GEO serves
one geographic area, and can theoretically cover about 41% of the earths surface. Companies proposing GEO
systems are planning on using between three and fifteen satellites to deliver worldwide service. The primary
advantage of GEO systems is that they are a proven technology. Most current communications satellites are
GEOs. A GEO system is also far less expensive than an LEO system, and also GEO ground stations can be
relatively simple because they need only target a fixed point relative to the earth. The main drawback for
GEO systems is called the latency factor. In order to obtain information from an Internet server, a signal has
to travel 35,000 km to the satellite, then 35,000 km back to the earth. This round trip takes approximately
one-quarter second.
LEOs orbit 20 times closer to the earth, between 700 km (450 miles) and 1350 km (700 miles) above the
earths surface. Each LEO is moving relative to the earth, covering a particular area for only a few seconds.
Because of this, a network of many satellites is required to cover the world.
Teledesic plans to launch a large number of LEOs that will be capable of offering high-speed Internet
access anywhere in the world. Service providers will include Teledesic, Globalstar, OrbComm, and SkyBridge
and M-Star (backed by Motorola). M-Star, while a broadband LEO system, is not aimed at the consumer market
like Teledesic; it is planned to offer high-bandwidth intercontinental links between network providers rather
than end users.
Satellites require a dedicated piece of the spectrum. Currently, the ITU has allocated 2.5 GHz of spectrum
for fixed satellite services in the 28 GHz Ka band. There are fourteen satellite applicants vying for pieces of
that 2.5 GHz. Requests for single applicants range from 750 MHz to the full 2.5 GHz, with most applicants
requesting 1 GHz of spectrum.
16
CABLE TELEVISION
The reason the Ka band has not been used in the past is that such high-frequency transmissions are
easily blocked. Buildings, trees, and other solid objects can cause a loss of signal. This makes these frequencies
unsuitable for use by ground-based systems, because they would require a large number of transmitters to
be able to avoid all obstacles. Satellites avoid many of the problems associated with blocking because their
signals come from directly overhead. Buildings and trees do not present an impediment to signals coming from
overhead satellites.
Power-Line Area Networks (Access via Utility Power Grids). There have been proposals lately
for traditional power companies to provide high-speed access via their existing grids. The biggest obstacle
to this technology is that data are scrambled when they pass through transformers. This obstacle is slowly
being overcome as several companies continue to work towards a solution. Nortel Networks has successfully
tested networks in Europe and Asia, where the transformer-to-customer ratio is 1/300. They have successfully
bypassed the transformers, but their speed has been limited, being comparable with cable modems and xDSL.
Data are transferred through the actual power wiring.
NIU (Network Interface Unit). NIU refers to high-speed connectivity through hybrid fiber-coax or
FTTH or FTTC networks using a network interface unit at the customer premises rather than an external or
PC-installed cable modem. Systems using NIUs usually support both data and telephony. Since the technology
can be utilized by cable operators, it is not actually a competitor to the cable industry but to cable modems.
Cable modem from DOCSIS, IEEE 802.14, IETF, DVB, and ATMF
EuroModem from DVB-RCC and ETS 300800
CABLE TELEVISION
17
EuroDOCSIS from DOCSIS (DOCSIS cable modem with some DVB technical compliance)
EuroBox
A number of cable operators from across Europe have developed the Eurobox Platform. This concept is
based on a common set-top box and a common application program interface API and conditional access
method. The box specification is used as a reference model by the cable industry and manufacturers.
Viaccess has been selected as the conditional access system, and Open TV as the API. The Eurobox
Platform has been successfully implemented, for example, in France, Sweden, and Denmark. However,
some cable operators, notably in the UK, do not appear to be following this platform in its entirety.
OpenCable
This initiative from CableLabs includes guidelines for building advanced set-top boxes including
feature enrichments to support broadband applications.
BIBLIOGRAPHY
1. G. Abe Residential Broadband, 2nd ed., Indianapolis, IN: Cisco Press, 2000.
2. K. Maxwell K. Maxwell Residential Broadband: An Insiders Guide to the Battle for the Last Mile, New York, Wiley:
1998.
3. B. C. Lindberg Digital Broadband Networks and Services, New York, McGraw-Hill: 1995.
4. C. A. Eldering N. Himayat F. M. Gardner CATV return path characterization for reliable communications, IEEE Commun. Mag., 33 (8): 6269 August 1995.
5. R. Rabbat K. Y. Siu QoS support for integrated services over CATV, IEEE Commun. Mag., 37 (1): 6468, January 1999.
Available www:http://www.comsoc.org/pubs/free/private/1999/jan/Rabbat.html
6. J. Karaoguz J. Yu V. Eyuboglu Comparison of single-carrier, multi-carrier and spread spectrum modulations for upstream
PHY layer in HFC CATV networks, IEEE 802.14a/98-018, San Diego, July 1998.
18
CABLE TELEVISION
INTERNET RESOURCES
Broadband Bob: http://catv.org
CableLAbs, research organization for US cable operators: http://www.cablelabs.com
CATV CyberLab: http://www.catv.org
DAVIC: http://www.davic.org
DVB/DAVIC Interoperability Consortium: http://www.dvb.org
DVB Web site: http://www.dvb.org
DOCSIS specifications: http://www.cablemodem.com
ECCA Web site: http://www.ecca.be
ETSI: http://www.etsi.org
FCC cable: www.fcc.gov
ITU: http://www.itu.org
MARYAN VAZQUEZ
-CASTRO
Carlos III University of Madrid
CHIRALITY
INTRODUCTION
Chiral media have the ability to discriminate between
left-handed and right-handed electromagnetic (EM) elds.
These media can be classied into two types: (1) isotropic
chiral media and (2) structurally chiral media. The
molecules of a naturally occurring isotropic chiral medium
are handed, while an articial isotropic chiral medium can
be made by randomly dispersing electrically small, handed
inclusions (such as springs) in an isotropic achiral host
medium. The molecules of a structurally chiral medium,
such as a chiral nematic liquid crystal, are randomly positioned but have helicoidal orientational order. Structurally
chiral media can also be articially fabricated either as
stacks of uniaxial laminae or using thin-lm technology.
Whereas considerable theoretical and experimental work
on isotropic chiral media has been reported at microwave
frequencies during the 1980s and the 1990s, microwave
research on structurally chiral media remains in an embryonic stage at the time of this writing (1). Therefore, the
major part of this article is devoted to isotropic chiral media.
NATURAL OPTICAL ACTIVITY
Ordinary sunlight is split into its spectral components by
a prism. A spectral component is monochromatic (i.e., it
has one and only one wavelength 0 in vacuum). The wavelength 0 of one of the visible spectral components lies anywhere between 400 nm (violet) and 700 nm (red). A spectral
component can be almost isolated from other spectral components by carefully passing sunlight through a series of
lters. Although ltering yields quasimonochromatic light,
many experiments have been and continue to be performed
and their results analyzed, assuming that the ltered light
is monochromatic.
Light is an EM wave with spectral components to which
our retinal pigments happen to be sensitive, and the consequent images, in turn, happen to be decipherable in our
brains. All optical phenomena can be generalized to other
electromagnetic spectral regimes.
Suppose that a monochromatic EM wave is propagating in a straight line in air, which is synonymous with vacuum (or free space) for our present purpose. Its electric
eld vector vibrates in some direction to which the propagation direction is perpendicular; the frequency of vibration is f=c/0 , where c=3 108 m/s is the speed of light in
vacuum. Its magnetic eld vector also vibrates with the
same frequency, but is always aligned perpendicular to the
electric eld vector as well as to the propagation direction.
Suppose that we x our attention on a certain plane that is
transverse to the propagation direction. On this plane, the
locus of the tip of the electric eld vector is the so-called
vibration ellipse, which is of the same shape as the locus
of the tip of the magnetic eld vector. A vibration ellipse is
shown in Fig. 1. Its shape is characterized by a tilt angle
as well as an axial ratio; in addition, it can be left-handed
if the tip of the electric eld vector rotates counterclockwise, or right-handed if otherwise. Similarly, an EM wave
is said to be elliptically polarized, in general; however, the
vibration ellipse can occasionally degenerate into a circle
(circular polarization) or even a straight line (linear polarization).
The shape of the vibration ellipse of monochromatic
light is altered after traversal through a certain thickness
of a so-called optically active medium. This phenomenon,
known as optical activity, was discovered around 1811
by F. Arago while experimenting with quartz. Crystals
are generally anisotropic, but J.-B. Biot observed around
1817 the optical activity of turpentine vapor, denitely an
isotropic medium. Isotropic organic substances were believed to have exclusively biological provenances, and in
1860 L. Pasteur argued that turpentine vapor exhibited
natural optical activity, but the optical activity of crystals
could not be similarly qualied. Pasteur was unduly restrictive. Isotropic optically active media, of biological or
other origin, are nowadays called isotropic chiral media,
because EM elds excited in them necessarily possess a
property called handedness (Greek cheir=hand). Facsimile
reproductions of several early papers are available (2).
CHIRAL MEDIA: NATURAL AND ARTIFICIAL
The molecules of an isotropic chiral medium are mirror
asymmetric (i.e., they are noncongruent with their mirror images). A chiral molecule and its mirror image are
called enantiomers (3). As examples, the two enantiomers
of 2-butanol are shown in Fig. 2. Enantiomers can have
different properties, although they contain identical atoms
in identical numbers. One enantiomer of the chiral compound thalidomide may be used to cure morning sickness,
during pregnancy, but its mirror image induces fetal malformation. Aspartame, a common articial sweetener, is
one of the four enantiomers of a dipeptide derivative. Of
these four, one (i.e., aspartame) is sweet, another is bitter, while the remaining two are tasteless. Of the approximately 1850 natural, semisynthetic, and synthetic drugs
marketed these days, no less than 1045 can exist as two or
more enantiomers; but only 570 were being marketed in the
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Chirality
2. P. Drude showed that chiral molecules can be modeled as spiral oscillators and theoretically veried a
rule Biot had given regarding OR spectra (6).
t) and B(r,
t) are the primitive or the fundamental
Thus, E(r,
EM elds, both functions of the three-dimensional position
vector r and time t; 0 =8.8541012 F/m and 0 =4107
H/m are, respectively, the permittivity and the permeability of vacuum; tot (r, t) is the electric charge density and
J tot (r, t) is the electric current density.
Equations (1a), (1b), (1c) and (1d) apply at any length
scale, whereas the charge and the current densities must
be specied not continuously but over a set of isolated
points. Electromagnetically speaking, matter is nothing
but a collection of discrete charged particles in vacuum.
As per the HeavisideLorentz procedure to get a macroscopic description of continuous matter, spatial averages
of all elds and sources are taken, while both tot (r, t) and
J tot (r, t) are partitioned into matter-derived and externally
impressed components. Then the Maxwell postulates at the
Chirality
Four constitutive property kernels appear in these equations; the dyadic -e(t) is the dielectric susceptibility kernel,
m(t)
and
are
called
the magnetoelectric kernels. Although a dyadic may be understood as a 33 matrix for
the purpose of this article, Chens textbook (10) is recommended for a simple introduction to the use of dyadics in
EM theory.
All four dyadic kernels in Eqs. (4a) and (4b) are causal
[i.e., -e(t) -0-0 for t0, etc.], because all materials must
exhibit delayed response. In addition, when we substitute
Eqs. (4a) and (4b) in Eqs. (2c) and (2d), respectively, a redundancy emerges with respect to Eqs. (2a) and (2b). Elimination of this redundancy leads to the constraint (11)
which has never been known to be violated by a physical material. Finally, crystallographic symmetries may also
impose additional constraints on the constitutive kernels.
A medium described by Eqs. (4a) and (4b) is said to be bianisotropic, since the constitutive kernels indicate anisotropy,
P(r,
t) and the magnetization M(r,
t)
Constitutive relations must be prescribed to relate the
in consequence of Eq. (5), where the scalar chi(t) is the chirality kernel. Equations (6a), (6b) describe the isotropic
chiral mediumthe most general, isotropic, linear electromagnetic material known to exist (12, 13).
Most commonly, EM analysis is carried out in the
frequency domain, not the time domain. Let all timedependent quantities be Fourier-transformed; thus
Chirality
or
where the asterisk denotes the complex conjugate. Equations (11a), (11b), (11c) and (11d) make it clear that while
() and () are true scalars, () has to be a pseudoscalar
since the numerator in either of its two denitions contains
a pseudovector. Other constitutive relationsequivalent
to Eqs. (9a) and (9b) and Eqs. (10a) and (10b)are also
used in the frequency-domain EM literature, but this article is restricted to the DBF constitutive relations Eqs. (10a)
and (10b), as they bring out the essence of chirality at the
very rst glance. An isotropic chiral medium and its mirror image share the same () and (), and their chirality
parameters differ only in sign.
Chirality
Here, Eexc (r , ) and Hexc (r , ) are the elds exciting the
particular inclusion; while and -hh () are the four linear
polarizability dyadics that depend on the frequency, the
constitution, and the dimensions of the inclusion. As the
inclusions are randomly oriented and any homogenizable
chunk of a composite medium contains a large number of
inclusions, -ee () and other terms in Eqs. (14) can be replaced by their orientationally averaged values. If the homogenized composite medium is isotropic chiral, this orientational averaging process must yield
The polarizability dyadics of electrically small, handed inclusions (e.g., springs) may be computed either with standard scattering methods such as the method of moments
(14) or using lumped-parameter circuit models (15). Provided that dissipation in the composite medium can be ignored, at a certain angular frequency, ee (), hh (), and
chi () are purely real-valued.
On applying the Maxwell Garnett homogenization approach, the constitutive relations of the homogenized composite medium (HCM) are estimated as follows (12):
where
Equivalently
Chirality
Figure 4. Optical rotation (OR) and circular dichroism (CD) spectra of a simple isotropic chiral medium. When the OR changes
sign, the CD records either a maximum or a minimum, which phenomenon is called the Cotton effect.
which are easier to analyze than Eqs. (18). The denominators on the left sides of Eqs. (20a) and (20b) suggest that
2 ()()2 () = 1 is not permissible for an isotropic chiral medium, as both wavenumbers must have nite magnitudes.
According to Eqs. (21a) and (21b), Q1 (r, ) and Q2 (r, )
are Beltrami elds (12). A Beltrami eld is parallel to its
own circulation. The concept arose early in the nineteenth
century, and has often been rediscovered. The easiest way
to think of a Beltrami eld is as a spiral staircase or a
tornado.
While Q1 (r, ) is a left-handed Beltrami eld, the negative sign on the right side of Eq. (15b) means that Q2 (r, )
is a right-handed Beltrami eld, because the two complexvalued wavenumbers 1 () and 2 () must have positive
real parts. Both wavenumbers also must have positive
imaginary parts in a causal material medium, since causal
materials must exhibit delayed response in the time domain and therefore must demonstrate EM loss (or attenuation) in the frequency domain.
As an isotropic chiral medium displays two distinct
wavenumbers at a specic frequency, it is birefringent.
More specically, because Q1 (r, ) and Q2 (r, ) have planewave representations possible only in terms of circularly
polarized plane waves, an isotropic chiral medium is often said to be circularly birefringent. The difference between 1 () and 2 () gives rise to natural optical activity.
While OR is proportional to the real part of [1 () 2 ()],
CD is proportional to the imaginary part of [1 () 2 ()].
The OR and CD spectra must be consistent with the
KramersKronig relations (16). The CD spectrum has a local maximum or minimum at the frequency where the sign
of the OR changes; this feature is labeled as the Cotton effect after H. Cotton, who reported it in 1895 (2). The OR
and CD spectra of a simple chiral medium are illustrated
in Fig. 4.
Solutions of Eqs. (23) in the Cartesian, the circular cylindrical, and the spherical coordinate systems are commonplace
(18).
Beltrami plane waves propagating in the +z direction
may be represented as
with A as the amplitudes, while x , y , and z are the Cartesian unit vectors.
In the circular cylindrical coordinate system (, , z) Beltrami elds with an exp(iz) dependence may be expressed
as the sums
are well behaved at = 0 with An and B n as the coefcients of expansion. The vector cylindrical wavefunctions
are given as
1/2
where = +( 2 2 ) ; ,
and z are the unit vectors in
the cylindrical coordinate system; J n () are the cylindrical Bessel functions of order n, and J n () are the respective rst derivatives with respect to the argument; while
Hn(1) () are the cylindrical Hankel functions of the rst
kind and order n, and Hn(1) () are the rst derivatives with
respect to the argument. For quasi-two-dimensional problems, = 0 because /z = 0 Parenthetically, in this paragraph denotes the radial distance in the xy plane and
should not be confused with the use of for charge density
elsewhere in this article.
Finally, with A smn and B smn as the coefcients of expansion, in the spherical coordinate system (r, , ), we have
Chirality
where
Chirality
wherein
which show clearly that the radiation eld of a point electric dipole in an isotropic chiral medium consists of lefthanded as well as right-handed components. If we have
instead a point magnetic dipole m located at the origin,
the source current densities are specied as J(r) = 0 and
Jm (r) = im(r) so that
Chirality
and 2 , is much smaller than unity, we may make the quasistatic approximation: W1 (r0 )
=W1 (r) and W2 (r0 )
=W2 (r)
for all r 0 V S . Then, Eqs. (44) simplify to
The Huygens principle allows the enunciation of the exterior surface equivalence principle. Consider a problem
in which surface Beltrami current densities W S 1 (r) and
W S 2 (r) exist on the exterior side of the surface S (see Fig.
6). As per Eqs. (37), these surface current densities act as
sources of the radiated elds
10
Chirality
is the relative permittivity dyadic in a reference plane designated as z=0. The rotation dyadic
Chirality
9.
10.
11.
12.
13.
Figure 8. Scanning electron micrograph of a 10-period chiral
sculptured thin lm made of silicon oxide. (From Professor Russell
Messier, Pennsylvania State University, with permission.)
14.
15.
16.
Although chiral STFs made of uorites, and singlefrequency OR measurements on them, were reported in
1959 (35), systematic experimental studiesalong with
scanning electron microscopic verication of the microstructural geometryappear to have begun only in
1995 (30). Figure 8 shows the scanning electron micrograph of a chiral STF made of silicon oxide. As typical
values of realized today range from 30 nm to 10 m,
microwave applications of these lms are yet not feasible,
but are likely to become an active area of research once
lms with 100 m become available. Many possible applications have been anticipated as the concept of STFs
for biological, optical, electronic, chemical, and other applications is beginning to take root, while many optical and
related applications have already been implemented (30,
31). Large-scale production appears feasible as well, with
adaptation of ion-thruster technology (36).
BIBLIOGRAPHY
1. O. N. Singh andA. Lakhtakia, eds., Electromagnetic Fields in
Unconventional Materials and Structures, Wiley, New York,
2000.
2. A. Lakhtakia, ed., Selected Papers on Natural Optical Activity,
SPIE Optical Engineering Press, Bellingham, WA, 1990.
3. J. Jacques, The Molecule and Its Double, McGraw-Hill, New
York, 1993.
4. B. Holmstedt, F. Hartmut, andB. Testa, eds., Chirality and
Biological Activity, Alan R. Liss, New York, 1990.
5. J. C. Bose, On the rotation of plane of polarisation of electric
waves by a twisted structure, Proc. Roy. Soc. Lond. 63: 146152
(1898) .
6. P. Drude, Lehrbuch der Optik, S. Hirzel, Leipzig, 1900.
7. K. F. Lindman, Uber
eine durch ein isotropes system von spiralformigen resonatoren erzeugte rotationspolarisation der
elektromagnetischen wellen, Ann. Phys. Leipzig. 63: 621644
(1920) .
8. R. Ro, Determination of the Electromagnetic Properties of
Chiral Composites, Using Normal Incidence Measurements,
17.
18.
19.
20.
21.
11
12
Chirality
34. A. Lakhtakia, Director-based theory for the optics of sculptured thin lms, Optik. 107: 5761 (1997) .
35. N. O. Young and J. Kowal, Optically active uorite lms, Nature. 183: 104105 (1959) .
36. M. W. Horn, M. D. Pickett, R. Messier, and A. Lakhtakia,
Blending of nanoscale and microscale in uniform largearea sculptured thin-lm architectures, Nanotechnology 15:
303310 (2004) .
AKHLESH LAKHTAKIA
Pennsylvania State University,
University Park, PA
202
CONFORMAL ANTENNAS
CONFORMAL ANTENNAS
A conformal antenna may be defined as an antenna whose
radiating aperture conforms to the surface of the body on
which it is mounted. Ideally, such antennas are flush
mounted or low profile (i.e., they do not protrude appreciably
out of the mounting surface). Basic slot and microstrip (patch)
antennas are typical examples of conformal antenna elements. The term conformal array has no unique definition.
Kummer (1) defines it as an array that is nonplanar. We shall
assume here that a conformal array consists of conformal (or
low-profile) antenna elements placed on a nonplanar surface.
The array surface is not generally at the disposal of the antenna designer and is often dictated by the specific application. For ground-based application, a conformal phased array
requiring coverage over 360 in azimuth (omnidirectional coverage) or coverage over a hemisphere the array surface may
be cylindrical or spherical, respectively. For conformal arrays
on aircraft, missiles, satellites, and surface ships, the array
shape may assume another form dictated by the contour of
the vehicle. Basic slot and microstrip antennas are extensively discussed in the literaturefor example, the textbook
by Balanis (2) is a typical reference. These antennas provide
ideal performance only when they are mounted on planar surfaces. During conformal application the curvature of the
mounting surface can affect their impedance and radiation
properties; such effects must be taken into account during the
design of such antennas.
The need for conformal phased arrays for aircraft and missile applications, and for ground-based arrays with omnidirectional coverage in azimuth or complete hemispherical coverage in space, has grown continually with requirements that
emphasize maximum utilization of available space and minimum cost. Many of the developments in conformal arrays
have been extensions of the concepts for planar phased
arrays, which are extensively discussed in the literature (for
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
CONFORMAL ANTENNAS
1.6
Slot
k =1.5
1.4
k = 2.0
Relative amplitude
1.2
r1 r2
1.0
k = 6.0
0.6
0.4
Exact solution
Creeping wave
Surface ray modes
0.2
0
Circumferential slots
b
a
or
Radial slots
a
k = 3.0
0.8
20
203
b
r3
40 60
80 100 120 140 160 180
Azimuth angle () degrees
Figure 1. Patterns of a thin axial slot on a perfectly conducting cylinder. (After Ref. 9.)
Figure 3. Slotted cone geometry.
Axial
polarization
f( ) = cos
Circumferential
polarization
f( ) = 1
Transition
zone
2
0.38 ( )1/3
k0a
g(0)/2~0.7
Figure 2. Approximate pattern of a thin slot on a conducting cylinder of radius a; k0 is the free-space propagation constant. (After Ref.
6.)
204
CONFORMAL ANTENNAS
80
z0
Metal cone
70
20 log10 S12 , dB
Tap-scattered ray
Circumferential slots
or
radial slots
z0
60
50
40
Frequency = 9.0 GHz
Theory
Measurements, z0 = 3.81 cm
30
20
Direct ray
10
30
60
90
120
150
180
210
0 , deg
2 0
Equivalent
cylinder
z0
Direct ray
Figure 4. Slotted cone and equivalent cylinder.
70
z0
50
40
45
30
Frequency = 9.0 GHz
Theory
Measurements, z0=10.16 cm
Measurements, z0=7.62 cm
Measurements, z0=5.08 cm
20
10
0
20 log10 S12 , dB
20 log10 S12 , dB
60
20
40
60
80
100
40
35
30
25
Cylinder calculations
Cone measurements
r1 = r2 = 45.53 cm
0 = 60.8 deg
c = 12.2 deg
120
0 , deg
Figure 5. Mutual coupling for circumferential slots on cylinder, 0
5.057 cm. (After Ref. 14.)
20
8.5
9.0
9.5
CONFORMAL ANTENNAS
20 log10 S12 , dB
90
r1 =
r3 =
0 =
c =
r 2 = 45.53 cm
92.39 cm
60.8 deg
12.2 deg
Cylinder calculations
Cone measurements
Cylinder calculations
QQ
;;
205
2b
z
a
80
1 1
70
60
8.5
9.0
9.5
Figure 8. Mutual coupling for circumferential slots versus frequency, 0 9.622 cm. (After Ref. 14.)
the TM10 mode, there is less radiation in the lower hemisphere for the TM01 mode. Wong and Ke (21) describe the design of this antenna for circular polarization by using the
TM01 and TM10 modes excited by a single coaxial feed located
on a diagonal line and the operating frequency chosen between the two lowest frequencies f 01 and f 10 given by Eq. (1).
Kashiwa, Onishi, and Fukai (22) describe the application
of a strip-line-fed cylindrically curved rectangular patch antenna as a small, portable antenna for mobile communication.
It has been found that near the resonant frequency the real
part of the input impedance approaches 50 . The radiation
patterns near the broadside direction are found to be similar
to those of the equivalent planar antenna; however, significant differences have been found in large off-broadside directions.
Radiation patterns of a cavity-backed microstrip patch antenna on a cylindrical body of arbitrary cross section have
been investigated theoretically and experimentally by Jin,
Berrie, Kipp, and Lee (23). The finite-element method has
been used to characterize the microstrip patch antennas, and
then the reciprocity theorem is applied in conjunction with a
two-dimensional method of moments to calculate the radiated
field. The method can be extended to characterize the radiation patterns of conformal microstrip patch antennas on general three-dimensional bodies.
f mn
c
=
2 r
m
2(a + h)1
2 n 2 1/2
+
(1)
2b
20 log10 S12 , dB
80
r 1 = 27.03 cm
r 2 = 25.88 cm
r 3 = 77.47 cm
0 = 80 deg
c = 11 deg
Cone calculations
Cylinder calculations
Cone measurements
70
60
50
8.2
8.4
8.6
8.8
9.0
9.2
9.4
The use of microstrip antennas on conical surfaces is of interest for aerospace vehicles with portions of their bodies conically shaped. Performance of a basic rectangular patch antenna on a metallic cone has been investigated theoretically
by Descardeci and Giarola (24). In the analysis the substrate
thickness is assumed to be very small compared with the distance of the patch to the cone apex, and the curvature radius
of the cone surface large compared with the operating wavelength. The capacitive effects and losses associated with surface wave have been neglected. Except for these assumptions,
the cavity model analysis used is general and applies to any
conical surface. Within the approximations made, the resonant frequency is not significantly affected by the conical surface. However, the radiation pattern is affected, with a conse-
206
CONFORMAL ANTENNAS
;;
;
;
;
;
yy;
;;
;;
yy
Input connector
Copperclad
CONFORMAL ANTENNAS
The Omni Microstrip Antenna
The omni microstrip or spiral slot antenna discussed in Refs.
2527 is essentially a short-circuited quarter-wavelength microstrip patch wrapped around a cylindrical surface to form a
spiral, as shown in Fig. 11. The cylinder is an epoxy fiberglass
dielectric, and the copper conduction are added using an electroless plating, masking, and electroplating technique. The
lower end and the inside of the patch are similarly plated to
form a short circuit and ground plane. The spiral slot antenna
has a height and diameter of 0.060 but, unlike conventional
small antennas, has well-matched input voltage standing
wave ratio (VSWR) of less than 2 : 1 over a 2% bandwidth at
238 MHz. The radiation patterns are similar to those of a
dipole oriented parallel to the cylinder axis, and the 1 dB
gain indicates an efficiency of better than 50%. The spiral slot
has also been developed for 42 MHz application in which the
antenna has to be contained in a 0.040 0.150 cylindrical
volume.
Dielectric-Filled Edge Slot Antenna
A class of circumferential slot antennas, called the DFES antennas, that are ideally suited for conformal mounting on conducting bodies of revolution has been described by Schaubert,
Jones, and Reggia (28). As shown in Fig. 12, the simplest form
of the antenna consists of a disk of dielectric substrate that is
copper coated on both sides and mounted between the two
halves of a conducting body so that the radiating aperture
coincides with the surface. The antenna is excited at the center by a coaxial line whose outer conductor is connected to the
lower conducting surface, and the inner conductor is extended
through the dielectric and finally connected to the conducting
surface at the upper end of the substrate. The input reflection
coefficients of the antenna are found to assume minimum values at some discrete frequencies, called the operating frequencies, where the antenna also radiates most efficiently.
The DFES antenna can be tuned for a desired operating frequency by using a number of axially oriented passive metallic
h
Inductive posts
plated through
Dielectric
(Teflon fiberglass)
Soldered
g/4
(a)
(b)
CONFORMAL ANTENNAS
207
Wrapped
on missle
15.2 cm
7.6 cm
0
0
+2
10 20 30
270
+2
dBi
30 20 10 90
270
Relative power
10 20 30
30 20 10 90
L = D
180
Feed
points
0
+2
;;;
Wrapped
to shape
dBi
Relative power
2 Elements
F = 660 MHz
180
+2
dBi
dBi
Microstrip
radiator
Feed
network
Input
270
30 20 10 90
10 20 30
Relative power
180
270
Relative power
180
2 Elements
F = 1210 MHz
0
+2
270
10 20 30
30 20 10 90
10 20 30
+2
dBi
30 20 10 90
Relative power
180
270
10 20 30
dBi
30 20 10 90
Relative power
180
4 Elements
F = 2000 MHz
= 0 / r
(2)
w = 0 = /2
2 r
(3)
L = D
(4)
L(r )1/2
LD =
0
(5)
NF > LD
(6)
90
30 2
(7)
320
40
310
50
300
60
350
340 10
330 20
30
10
20
30
40
50
60
290
70
70
280
80
270
90
80
90
100
0 10
270
90
30 2
100
110
0 10
270
110
120
120
130
130
140
180
Elevation
pattern
E ( ) +0
180
F = 6330 MHz
Azimuthal
pattern
E ( ) 90
140
150
160
160
150
Figure 16. Measured E-plane pattern of the 8 in. (203 mm) wraparound microstrip antenna. The antenna pattern is a figure of revolution about the missile axis. (After Ref. 31.)
208
CONFORMAL ANTENNAS
1. Radiation patterns are almost independent of the pressure of the patch when the radius of the sphere is much
larger than the strip width.
2. The larger the radius of the sphere (a 0), the better
the omnidirectional pattern.
3. The dielectric constant r does not have significant influence on the pattern shape. The radiation intensity
tends to increase with increase of the dielectric constant.
4. The shape of the radiation patterns remains almost unchanged for different substrate thickness (h) for h 0.
However, sidelobe levels increase with increase of r.
Radiation patterns of rectangular microstrip patches arrayed circumferentially on a circular cylinder (wraparound
array) have been computed in Ref. 34. Both axial and circumferential patches, using axial and circumferential modes of
excitation, respectively, have been used. In general, it has
been found that the number of circumferential patches required for a given ripple in the gain pattern is considerably
less than that required in the axial case, thus simplifying the
feed network for the former case. Results given in Ref. 34
compared favorably with reported measurements.
A Patch Array for Aircraft
A patch array designed for an aircraft to satellite communication link is described by Sanford (35) and is shown in Fig. 17.
Eight patches are mounted together with the phase shifting
and feeding circuitry to scan the beam in the elevation direction. Designed for operation at 1.5 GHz, the array, including
radome, is 3.6 mm thick. Element phasing was optimized for
maximum multipath rejection at low scan angles and to account for the curvature of the mounting surface. Each element was pointed in a different direction and has an inherent
phase error relative to the center elements. A digital computer was used to determine how the design parameters actually affect the performance of the array. The spacing of the
array elements must be greater than 0.32 (in free space) because the physical size of the radiating element on teflon fiberglass requires thin space. The spacing required to prevent
Phase shifting
circurity
Patch
radiators
Aircraft
body
Figure 17. Conformal array for aircraft application. (After Ref. 35.)
1 + sin
(8)
where D is the separation distance between the patch element, is the wavelength in free space, and is the maximum beam steering angle. For a 50 maximum steering
angle, D must be less than 0.57.
Concentric Microstrip Ring Arrays
Bhattacharryya and Garg (36) describe the design of a concentric annular ring microstrip antenna array that can be excited by means of a single feed by interconnecting two consecutive rings with an impedance transformer. The feasibility of
such an antenna is based on the observation that annular
rings with different mean radii can be designed to resonate
at the same frequency for the TM12 mode. An impedance
bandwidth of about 5% for VSWR 2 has been reported in
Ref. 36. It has been found possible to control the principal
plane patterns for concentric arrays independently of each
other by appropriately designing the feed system.
Saha-Misra and Chowdhury (37) describe electromagnetically fed concentric microstrip ring arrays using the log periodic principle that have been reported to have increased impedance and radiation pattern bandwidths. Specifically,
circular, square, and triangular concentric rings have been
investigated. Generally, these antennas work at multiple
bands of frequencies with some bands having larger bandwidths than standard microstrip antennas. With a nonuniformly spaced concentric annular ring array, almost 20%
bandwidth for VSWR 2 has been reported. A planar, wideband feed for a slot spiral antenna has been described by
Nurnberger and Volakis (38). The antenna has been developed for operation at very high frequency (VHF) frequencies.
In contrast to most traditional printed spiral antenna designs, the one reported in Ref. 38 incorporates a completely
planar spiral microstrip balun feed, thereby making it attractive for a variety of conformal applications.
CONFORMAL ARRAYS
Antenna arrays conforming to a nonplanar surface are suitable and may even be a requirement for a number of applications. For example, phased arrays of flush-mounted elements
conformally mounted on the surface of an aircraft or missile
reduce the aerodynamic drag and hence are preferable. Also,
in some cases a nonplanar array surface may provide some
natural advantage for broad-beam coverage in space. Spherical, cylindrical, and conical arrays have been developed for
ground, airborne, and missile applications. We shall consider
here the class of conformal arrays where the radiating surface
is nonplanar with a radius of curvature large compared to
the operating wavelength. Conformal arrays that are highly
curved are generally difficult to design because of the following reason (1,3,8,4):
1. Array elements point in different directions and so it is
often necessary to switch off those elements that radiate
primarily away from the desired direction of radiation.
CONFORMAL ANTENNAS
209
Cylindrical Arrays
Conformal elements like microstrips and slots arrayed around
the circumference of large metal cylinders have been used to
obtain omnidirectional pattern coverage. Such coverage may
also be obtained with the help of wraparound antenna, as discussed earlier. References 4246 show that an array of slots
equally spaced around the circumference of a cylinder can
produce a pattern with very low ripple. Croswell and Knop
(43) have obtained extensive numerical data using realistic
patterns for slots on perfectly conducting planes. In such
arrays the design parameters are the numbers of elements,
radiating elements, and feed network. The number of elements is chosen to provide a nearly omnidirectional pattern;
the minimum number of elements is decided by the allowed
amplitude ripple. The evaluation of the amplitude ripple can
be given in terms of the fluctuation, which is defined as the
ratio of maximum F to minimum F, where F is the total farfield pattern of an S-element circular array and is given by
(42)
F =S
N
An ( j)n
n=0
dn
[J (z) + 2( j)SJS (z) cos S]
dz n 0
(9)
An cosn
(10)
n=0
f ( ) =
(1 + cos /2)/2
(2 + 3 cos + cos 2 )/6
(11)
Pattern fluctuations as a function of size of cylinder and number of elements for the preceding two single-element patterns
are given in Ref. 43. Cylindrical phased arrays, where selected sections are illuminated to provide a beam in a certain
direction, are sometimes found advantageous to use for some
requirements. Sophisticated types of electronic switches for
such circular arrays are based on a concept originally proposed by Shelton (45) and developed by Sheleg (46). The antenna uses a Butler matrix-fed circular array with fixed phase
shifters to execute current modes around the array and variable phase shifters to provide continuous scanning of the radiated beam over 360. The operation was experimentally demonstrated with a 32-dipole circular array.
The principles involved in scanning a multimode array are
readily seen by considering a continuous distribution of current, as described by Sheleg (47). Figure 19 shows the configuration of a continuous cylindrical sheet of vertical current
elements around a vertical conducting cylinder of radius a.
Referring to Fig. 19, consider a current distribution I() to be
the sum of a finite number of continuous current modes Ine jn
with N n N. The radiation pattern for I() Ine jn is
210
CONFORMAL ANTENNAS
then given by
E() =
N
Cn ein
(12)
n=N
Equal line
lengths
2a
In =
2K j n Jn
Butler matrix
(13)
2a
Variable phase
shifters
Fixed phase
N shifters
N1
Power divider
(14)
Figure 20. Schematic diagram of scanning multimode array. (After
Ref. 46.)
E() =
N
2N + 1
e jn = sin
n=N
2
sin
(15)
(16)
J = N1 x
J=N x
= 0
J=1
x
J=2
BIBLIOGRAPHY
1. W. H. Kummer, Preface, IEEE Trans. Antennas Propag., AP-22:
13, 1974.
2. C. A. Balanis, Antenna Theory, 2nd ed., New York: Wiley, 1997.
CONFORMAL ANTENNAS
3. R. C. Johnson and H. Jasik, (eds.), Antenna Engineering Handbook, 2nd ed., New York: McGraw-Hill, 1984.
4. R. C. Hansen, Significant Phased Array Papers, Dedham, MA:
Artech House, 1973.
5. Y. T. Lo and S. W. Lee, (eds.), Antenna Handbook: Theory, Applications, and Design, New York: Van Nostrand Reinhold, 1988.
6. R. J. Mailloux, Phased Array Antenna Handbook, Boston: Artech
House, 1994.
7. W. H. Kummer, guest editor, Special Issue on Conformal Arrays,
IEEE Trans. Antennas Propag., AP-22: 1150, 1974.
8. R. C. Hansen, Microwave Scanning Antennas, Vols. 1-III, New
York: Academic Press, 1964.
9. P. H. Pathak and R. G. Kouyoumjian, An analysis of the radiation from aperture in curved surfaces by the geometrical theory
of diffraction, Proc. IEEE, 62: 14381447, 1974.
10. P. H. Pathak et al., A uniform GTD solution for the radiation
from sources on a curvex surface, IEEE Trans. Antennas Propag.
AP-29: 602622, 1981.
11. W. D. Burnside, R. J. Marhefka, and C. L. Yu, Roll plane analysis
of on-aircraft antennas, IEEE Trans. Antennas Propag., AP-21:
780786, 1973.
12. W. D. Burnside et al., A Study of KC-135 aircraft antenna patterns, IEEE Trans. Antennas Propag., AP-23: 309316, 1975.
13. R. J. Mailloux, Conformal and low-profile arrays, in Ref. 3.
14. K. E. Golden, G. E. Stewart, and D. C. Pridmore-Brown, Approximation techniques for the mutual admittance of slot antennas on
metallic cones, IEEE Trans. Antennas Propag., AP-22: 4348,
1974.
15. K. E. Golden and G. F. Stewart, Self and mutual admittance of
rectangular slot antennas in the presence of inhomogeneous
plasma layer, IEEE Trans. Antennas Propag., AP-17: 763771,
1969.
16. D. C. Chang, (ed.), IEEE Trans. Antennas Propag., AP-29: 1
182, 1981.
17. I. J. Bahl and P. Bhartia, Microstrip Antennas, Norwood, MA:
Artech House, 1980.
18. C. M. Krowne, Cylindrical-rectangular microstrip antenna, IEEE
Trans. Antennas Propag., AP-31: 194199, 1983.
19. K.-M. Luk, K.-F. Lee, and J. S. Dahele, Analysis of the cylindrical-rectangular patch antenna, IEEE Trans. Antennas Propag.,
AP-37: 143147, 1989.
20. K. R. Carver and J. W. Mink, Microstrip antenna technology,
IEEE Trans. Antennas Propag., AP-29: 224, 1981.
21. K.-L. Wong and S.-Y. Ke, Cylindrical-rectangular microstrip for
circular polarization, IEEE Trans. Antennas Propag., AP-41:
246249, 1993.
22. T. Kashiwa, T. Onishi, and I. Fukai, Analysis of microstrip antennas on a curved surface using the conformal grids FD-TD method,
IEEE Trans. Antennas Propag., AP-42: 423427, 1994.
23. J. M. Jin et al., Calculation of radiation patterns of microstrip
antennas on cylindrical bodies of arbitrary cross section, IEEE
Trans. Propag., AP-45: 126132, 1997.
24. J. R. Descardeci and A. J. Giarola, Microstrip antenna on a conical surface, IEEE Trans. Antennas Propag., AP-40: 460463,
1992.
25. H. S. Jones, Design of dielectric-loaded circumferential slot antennas of arbitrary size for conical and cylindrical bodies, Report
HDL-TR-1684, Hary Diamond Laboratories, Adelphi, MD, 1974.
26. D. H. Schaubert, A. R. Sindons, and F. G. Farrar, The spiral slot:
a unique microstrip antenna, Proc. 1978 Antenna Applic. Symp.,
Univ. of Illinois Allerton Conf., Hary Diamond Laboratories, October 1978.
211
DIPAK L. SENGUPTA
University of Detroit Mercy
CONICAL ANTENNAS
221
P(r, , )
CONICAL ANTENNAS
1
H e j r
H =
r sin 0
(1)
E = H
(2)
where 2/ (: wavelength) is the free-space wave number and 0 / 0 is the free-space wave impedance. The
characteristic impedance K of the biconical transmission line
is given by the ratio of the transmission voltage (i.e., the integral of E along the cone-meridian) to the conduction current
K = ln cot
(3)
800
Characteristic impedance K ( )
This article describes characteristics such as input impedance, radiation pattern, and directivity gain of conical antennas consisting of solid conducting cones, conducting conical
plates, or their modifications. Conical conducting structures
on which simple antenna elements such as dipole and a slot
are mounted are also described. An important general feature
of conical antennas is their lack of sensitivity to frequency
variation, that is, their broadband characteristics. Note that
a conical horn antenna is described in HORN ANTENNAS.
The history of the practical use of conical antennas is long.
Sir Oliver Lodge constructed a biconical antenna in 1897 and
made a wireless communication experiment, while a single
cone antenna on the ground and a fan (flat triangular) antenna were used by Marconi and others. The history of the
theory of conical antennas is also long. The spherical coordinate is one of the few coordinates for which the method of
variable separation in electromagnetic field problems can be
applied. The conical surface is defined by a constant polar
angle in the spherical coordinate system. For this reason conical antennas and biconical antennas, in particular,
have been extensively investigated by Schelkunoff (1,2),
Smith (3), Tai (4,5), Papas and King (6), and many others.
In the limit as 0 and , the biconical antenna is reduced to a vanishingly thin linear antenna that is very sensitive to frequency (that is, of narrow bandwidth). As the cone
angle increases, the antenna shows broadband characteristics
that make it useful for practical applications.
Figure 1 shows an infinitely long, symmetric biconical conductor with a half-cone angle and an infinitesimally small
feeding gap ( gap). The antenna is assumed to be located in
free space. This structure can support the transverse electromagnetic (TEM) transmission line mode, that is, the outwardpropagating principal spherical wave mode expressed by
700
600
K = 120 In cot
500
400
300
200
100
0
10
20
30
40
50
60
Half-cone angle (deg)
70
80
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
222
CONICAL ANTENNAS
400
2 = 5
Resistance ( )
300
2
h
10
200
20
30
40
100
40 50 60
90
30
20
5
50
60
70
80
90
70
10
20
5
30
40
50
60
80
90
0
0
Figure 3. Symmetric biconical antenna.
50
100
150
200
Antenna length h (deg)
250
300
driven by a coaxial line has been numerically analyzed by using the finite difference time domain method (9).
When the upper half-cone of the biconical antenna is
mounted on an infinite conducting plane (ground plane), the
antenna forms a conical unipole having one half of the input
impedance of the biconical antenna. Figures 5 and 6 [Brown
and Woodward, Jr. (10)] show respectively the measured input resistance and reactance of a conical unipole having flat
caps instead of spherical caps. It is clear that the antenna
tends to have a constant input resistance and a small reactance around zero versus frequency, showing broadband
characteristics as the cone-angle is increased.
The radiation pattern of the biconical antenna has been
computed by Papas and King (6) and by Bevensee (11). Figure
7 shows the far-zone electric field pattern (6) for the cone
angle of 30. It is found that the patterns are not much
different from those of a straight wire antenna.
Theoretical analysis of biconical antennas loaded with
and/or immersed in dielectric, lossy, and ferromagnetic mate200
2
h
100
Resistance ( )
2 = 5
10
20
30
40
50
60
90
90
90
70
80
60
50
70 40
80
60
50
40
30
20 10
30
5 90
60
50
40
20
30
100
10
20
10
5
200
Yi
Yt
a
Figure 4. Equivalent circuit of a symmetric biconical antenna.
300
50
100
150
200
Antenna length h (deg)
250
300
CONICAL ANTENNAS
60
60
ka = 2
ka = 1.5
ka 1
(a)
223
70
70
80
80
90
(a)
90
(b)
60
ka = 3
ka = 2.5
60
ka = 4
ka = 3.5
70
80
90
(c)
70
80
90
(d)
60
60
ka = 6
ka = 5.5
ka = 5
ka = 4.5
70
70
80
80
90
(e)
rials has been provided by Schelkunoff (2), Tai (4), Polk (12),
and others. These topics are reviewed by Wait (13). The theory of an asymmetric biconical antenna was also discussed by
Schelkunoff (2). The variational approach by Tai was extended to a semi-infinite asymmetric conical antenna consisting of an infinite cone and a finite cone (14).
To reduce wind resistance and/or weight, a solid biconical
antenna can be replaced by a skeletal conducting wire structure using several radial rods (15,16). It has been found, however, by analysis using the moment method that a considerable number of wires (e.g., 16) is required to approximate the
solid biconical antenna.
ANTENNAS ON CONICAL STRUCTURE
In practice, a conical structure on which simple antenna elements such as dipole, disk, cone, slot, or patch are mounted
to excite a cone is often used. The infinite cone excited with
an axial dipole at the tip has shown (14) that the strong radiation occurs along the small-angle cone unless the dipole
length is about a half-wavelength. Figure 8 shows a finite
wide-angle cone excited with a quarter-wavelength long and
a half-wavelength long dipole (17). Note that the maximum
radiation can be directed toward the horizontal plane by
proper choice of a and 0 at a desired frequency. A cone excited with an axially symmetric circumferential slot close to
the tip (18) shows radiation characteristics similar to those of
a dipole-excited cone, since a small circumferential slot (mag-
90
Figure 7. Far-zone electric field patterns of a biconical antenna, 30. Patterns do not change
very much from ka 0 to ka .
netic current loop) is equivalent to a small axial electric dipole. When the circumferential slot is not too close to the tip,
that is, apart by 2.5 wavelength (2 30), the radiation pattern shows a rather complicated lobe structure (18). The cone
excited with a circular disk at the tip is called a discone antenna (19). This antenna is fed with a coaxial cable whose
inner conductor terminates on the center of the disk and
whose outer conductor terminates on the tip of the cone. The
radiation pattern of the discone is similar to that of a dipole
antenna, but its input impedance bandwidth is exceedingly
broad compared with an ordinary dipole antenna.
Figure 9 shows the measured input VSWR (for 57.6 cable) versus frequency of the finite conical antenna with different exciting elements at its tip (17). The cone excited with a
conical element (an asymmetric biconical antenna) indicates
very broadband characteristics, particularly in lower-frequency regions, without affecting the radiation pattern.
TRIANGULAR (BOW-TIE) ANTENNAS
A triangular plate antenna above a conducting ground plane
and a bow-tie antenna are shown in Figs. 10(a) and 10(b).
These antennas also possess broadband characteristics,
though not as broad as a solid conical antenna. The theoretical characteristics of the bow-tie antenna have been obtained
numerically (20) by using the method of finite difference time
domain (FDTD). Figures 11(a) and 11(b) show the calculated
input impedance. The input impedance of the triangular plate
224
CONICAL ANTENNAS
o
a
2
2
(a)
h
2h
(b)
(a)
(b)
Figure 10. (a) Triangular plate antenna and (b) bow-tie antenna.
(c)
Figure 8. Far-zone electric field pattern of a tip-excited conical antenna, a 50. (a) 0 90, l /4; (b) 0 103.8, l /4; (c) 0
103.8, l /2. Radiation beam can be directed to horizon by slanting
a cone downward (from 17).
7
0.635 cm
6
Thin
polystyrene
disk
VSWR
5
0.238 cm
L = 7.45 cm
L = 0.835 cm
0.90 cm
0.835 cm
0.238 cm
2
Figure 9. VSWR versus frequency of a
disk and a cone excited with a cylindrical
element or a conical element showing that
a cone excited with a conical element is
most broadband (from 17).
Frequency (GHz)
10
11
12
13
CONICAL ANTENNAS
225
90
60
120
0.8
0.6
30
700
= 15
= 45
600
0.4
= 30
= 60
0.2
Resistance ( )
500
0
400
300
200
210
330
100
0
0.5
1
1.5
Length of bow-tie antenna 2h ( )
2h = 0.5
2h =
2h = 1.5
(a)
400
= 15
= 45
Resistance ( )
200
300
270
= 30
= 60
30
30
0.8
0.6
60
0
0.4
200
0.2
90
400
600
0
0.5
1
1.5
Length of bow-tie antenna 2h ( )
(b)
Figure 11. Input impedance of a bow-tie antenna showing the broadband characteristics with increasing flare angle (from 20).
120
2h = 0.5
2h =
2h = 1.5
150
180
(b) x-z plane
226
= 15
= 45
= 30
= 60
4
2
0
2
4
0.2
0.2
0.2
0.2
0.2
0.2
0.2
Length of bow-tie antenna 2h ()
0.2
antenna above the ground plane is half of that of the bowtie antenna.
The far-zone electric field patterns in the x-y plane and in
the x-z plane are shown in Figs. 12(a) and 12(b), respectively.
Note that the radiation is enhanced in the direction perpendicular to the antenna plate for the antenna length 2h ,
because the radiation from the antenna surface current is
added in phase in that direction. The theoretical directivity
gain of the bow-tie antenna in the direction of the x-axis is
shown in dBi in Fig. 13 versus the antenna length 2h/ for
various cone angles (20). It is noted here that the bow-tie antenna can also be simulated by several radial wire rods as the
solid biconical antenna.
BIBLIOGRAPHY
1. S. A. Schelkunoff, Principal and complementary waves in antennas. Proc. IRE, 34 (1): 2332, 1946.
2. S. A. Schelkunoff, Advanced Antenna Theory, New York: Wiley,
1952.
3. P. D. P. Smith, The conical dipole of wide angle, J. Appl. Phys.,
19 (1): 1123, 1948.
4. C. T. Tai, On the theory of biconical antennas, J. Appl. Phys., 19
(12): 11551160, 1948.
5. C. T. Tai, Application of variational principle to biconical antennas, J. Appl. Phys., 20 (11): 10761084, 1949.
6. C. H. Papas and R. W. P. King, Radiation from wide-angle conical
antennas fed by a coaxial line, Proc. IRE, 39 (1): 4951, 1951.
7. S. A. Saoudy and M. Hamid, Input admittance of a biconical antenna with wide feed gap, IEEE Trans. Antennas Propag. 38 (11):
17841790, 1990.
8. V. Badii, K. Tomiyama, and D. M. Grimes, Biconical transmitting
antennas, a numerical analysis, Appl. Comput. Electromagn. Soc.
J. 5 (1): 6293, 1990.
9. J. G. Maloney, G. S. Smith, and W. R. Scott, Jr., Accurate computation of the radiation from simple antennas using the finite-difference time-domain method, IEEE Trans. Antennas Propag., 38
(7): 10591068, 1990.
10. G. H. Brown and O. M. Woodward, Jr., Experimentally determined radiation characteristics of conical and triangular antennas, RCA Rev., 13 (4): 425452, 1952.
11. R. M. Bevensee, Handbook of Conical Antennas and Scatterers,
New York: Gordon and Breach, 1973.
SABURO ADACHI
Tohoku Institute of Technology
DIELECTRIC-LOADED ANTENNAS
279
DIELECTRIC-LOADED ANTENNAS
A transmit antenna converts the energy of a guided wave in
a transmission line into the radiated wave in an unbounded
medium. The receive antenna does the reverse. The transmission lines such as waveguides, coaxial lines, and microstrip
lines use conductors mostly to confine and guide the energy,
but antennas use them to radiate it. Because the radiated
energy is in an unbounded region, phase control is often used
to direct the radiation in the desired direction. Dielectrics
play an important role in this process, and this article discusses a few representative cases. An important antenna parameter is its directivity, which is the measure of its control
over the energy flow. To increase the directivity the antenna
size must be increased, and the influence of dielectrics on
their performance changes considerably. Thus, in this article,
the use of dielectrics in antenna applications is divided into
two categories: large high-gain antenna applications and
small low-gain antenna applications.
In high-gain antenna applications, reflectors and lenses
are used extensively (1). They operate principally on the basis
of their geometry. Consequently, they are relatively low cost,
reliable, and wideband. Reflectors are usually made of good
conductors, and thus have lower loss, and because of their
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
280
DIELECTRIC-LOADED ANTENNAS
high strength they can be made light. However, reflectors suffer from limited scan capability. Lenses, on the other hand,
because of transparency, have more degrees of freedomthat
is, two reflecting surfaces and the relative permittivity or refractive index. They also do not suffer from aperture blockage.
However, lenses have disadvantages in large volume and
weight.
In microwave antenna applications, lenses have numerous
and diverse applications, but in most cases they are large
with respect to the wavelength. Thus, physical and geometrical optics apply, and most of the lens design techniques can be
adopted from optics to microwave applications. The aperture
theory and synthesis techniques can also be used effectively
to facilitate designs. In addition, the use of optical ray path
in lens design makes the solution frequency-independent. In
practice, however, the lens size in microwave frequencies is
finite with respect to the wavelength, and the feed antenna is
frequency-sensitive. Thus, the performance of the lens antenna also becomes frequency-dependent.
Natural dielectrics at microwave frequencies have reflective indices larger than unity, and for collimation they need
convex surfaces. However, artificial media using guiding
structures, such as waveguides, are equivalent to dielectrics
with refractive index less than unity, and they result in concave lenses. They are usually dispersive, resulting in variation of the refractive index with frequency, and have narrower operating bandwidths.
In small antennas, dielectrics are used often to improve
the radiation efficiency and polarization of other antennas,
such as waveguides and horns. This is important in telecommunication applications where polarization control is required to implement frequency reuse and minimize interference, especially in satellite and wireless communications.
Horn antennas and reflector feeds are examples that incorporate dielectrics or lens loading to improve performance (2).
Another area of important dielectric use is insulated antennas, in biological applications and remote sensing with
buried or submerged antennas. The use of dielectric loading
eliminates direct radio-frequency (RF) energy leak into the
lossy environments, and it ensures radiative coupling into the
target objects. Often a full wave analysis is needed to provide
a proper understanding of resonance property and coupling
mechanism to the surrounding media.
Finally, the antenna miniaturization depends primarily on
the dielectric loading. Low-loss dielectrics with medium to
high relative permittivities are now available and are used
increasingly to reduce the antenna size. A number of important areas include dielectric-loaded waveguides and horns, dielectric resonator antennas, and microstrip antennas. By aperture loading of waveguides and small horns, excellent
pattern symmetry and low cross-polarization can be obtained,
which are essential features of reflector and lens feeds. In addition, the dielectric loading reduces their size and makes
them useful candidates for multiple beam applications, using
reflectors and lenses. Miniaturization of the antenna is also
an important requirement in wireless communications. Microstrip patch or slot antennas with high relative permittivity
substrates play an important role in this area, and their derivatives are used in most applications.
DIELECTRIC LENS ANTENNAS
In optical terms a lens produces an image of a source point
at the image point. These points could be located anywhere
Feed antenna
n>1
Concave lens
n = r
Feed antenna
n>1
Concave lens
Figure 1. Geometry of lens antennas.
DIELECTRIC-LOADED ANTENNAS
Phase
front
y
P1(x1, y1)
(r1, 1)
r3
P2(x2, y2)
(r2, 2)
L1
P3
r1
r2
2
L0
A
(x3, 0)
281
dr1
nr1 r2 sin(2 1 )
=
d1
r3 n[r2 cos(2 1 ) r1 ]
(7)
dr2
nr1 r2 sin(2 1 ) + r2 r3 sin 2
=
d2
r3 sin 2 n[r2 r1 cos(2 1 )]
(8)
and
F
S1
S2
x1 = r1 cos 1
reference, and its length from S to C is compared with that of
the ray passing through points P1, P2, and P3. This provides
the following equation:
SP1 + nP1 P2 + P2 P3 = SA + nAB + BC
(1)
or
r1 + nr3 + L1 = F + nT + L0
(2)
L1 = (x3 x2 )
(3)
L0 = x3 (F + T )
and F and T are the lens focal length and axial thickness and
are therefore constant lengths defining the lens.
Enforcing Fermats principle at points P1 and P2 results in
differentiation of the path length in Eq. (1) in terms of its
variables x1, y1 and x2, y2 and setting it to zero. This provides
the slope of the lens surface profiles at each point P1 and
P2.
At point P1, one obtains
d
d
d
[r + nr3 + L1 ] =
[F + nT + L0 ] =
L
dx1 1
dx1
dx1 0
(4)
(5)
x2 = r2 cos 2
(9)
y2 = r2 sin 2
r3 = |r1 r2 | = [r21 + r22 2r1 r2 cos(2 1 )]1/2
Solutions of Eqs. (7) and (8) give the lens profiles in polar
coordinates, which are often more compact in form. Also, for
some simple lens configurations they result in well-known
and easily recognizable parametric equations of the conic sections, generalizing the solution.
Examples, Simple Lenses
y1 = r1 sin 1
(6)
The lens design becomes considerably easier if one of its surfaces is predetermined. This eliminates one of the differential
equations, as the surface profile is already known. The planar
and spherical surfaces are among the simpler surfaces to select. The planar surfaces are normal to the lens axis. Such
selections give simple profile equations. The planar surface is
described by a constant x coordinate, and the spherical one is
described by a constant polar coordinate r. These simplifications also assist in solutions of the other lens profile, for
which an analytic solution can also be determined. Since either of the lens profiles can be predetermined as planar or
spherical, four possible solutions exist. Only two, however, result in simple conical sections.
If the second surface S2 is assumed to be planar and normal to the lens axis, the rays arriving from the right-hand
side, parallel to the lens axis x, enter the lens unaffected and
change direction only after the first lens surface S1. Then they
focus at S. That is, only the S1 surface of the lens collimates
the beam. Looking from the left side, spherical rays originating from the focal point S enter the lens S1 and become parallel to its axis. Thus, after leaving the lens at S2, since they
are normal to S2, their direction remains unchanged. In this
case, the active surface S1 of the lens is a hyperbola in a cylindrical lens but is hyperboloid in a rotationally symmetric lens.
If the surface S1 is spherical, it becomes inactive, since the
focal point is a point source and rays emanating from it constitute spherical waves. Thus, when S1 is predetermined as a
282
DIELECTRIC-LOADED ANTENNAS
and
1 max = cos1
P1(x1, y1)
(r1, 1)
r1
= tan1
1
n
D
2
(16)
(F + T )
Equation (16) shows that, for a given dielectric, the lens aperture angular size is limited by its refractive index n. In other
words, with common dielectrics there is a limit on the compactness of the lens. That is, the focal length F cannot be
reduced beyond the limit specified by Eq. (16).
S
F
T
S1
S2
Lens with Planar S1. In this case, both lens surfaces contribute to the beam collimation. Its surface can be determined
similar to case (a) by enforcing x1 F and infinite slope for
S1 (Fig. 4). The results are (3)
Hyperboloid
Figure 3. Geometry of lens with a planar surface S2.
x1 = F
spherical surface, they enter the lens unaffected. Their collimation is done entirely by the lenss second surface S2. Its
surface is again a conic section and its cross section is elliptic.
In the other two cases, both surfaces S1 and S2 of the lens
participate in beam collimation and consequently are enterdependent and more complex.
T=
x2 = F + T
(10)
y2 = y1
(11)
1
(n 1)1 [(4F 2 + D2 )1/2 2F]
2
(12)
y2 =
(13)
They represent rectangular and polar equations of a hyperbola, which is the lens profile on S1. They can be used also to
determine the lens thickness on the axis. For this, one can
use two extreme rays passing through its tip and the axis.
The equality of the electrical lengths gives
F + nT = r1 (1 max ) =
D
2
2
x2 + (n 1)R
n
+ (F + T )
(14)
P1(F, y1)
(r1, 1)
T = (n + 1)1
(n + 1)D
+ F2
4(n 1)
2
1/2
x2
P2(x2, y2)
(r2, 2)
x
D
T
S1
1/2
2
1
F
(18)
1/2
2
(n 1)R
n cos 2
(17)
S2
(15)
Figure 4. Geometry of lens with a planar surface S1.
DIELECTRIC-LOADED ANTENNAS
283
y
Spherical
P1(n1, y1)
(F, 1)
1
A()
=
A(1 )
F
P2(x2, y2)
(r2, 2)
(n cos 1 )3
(n 1)2 (n cos 1 1)
1/2
(22)
1
A()
=
A(1 )
F
(n cos 1 1)3
(n 1)2 (n cos 1 )
1/2
(21)
284
DIELECTRIC-LOADED ANTENNAS
Table 1. Amplitude Distributions for the Hyperbolic and Elliptic Lenses of Figs. 3 and 5
Amplitude Ratio,
A()
A(1)
Hyperbolic lens equation
[Eq. (21)]
Elliptic lens equation
[Eq. (22)]
Relative value
dB
Relative value
dB
10
20
30
35
40
45
50
1.0
0.0
1.0
0.0
0.928
0.65
1.060
0.51
0.733
2.70
1.26
2.01
0.466
6.64
1.69
4.55
0.328
9.70
2.06
6.29
0.196
14.17
2.67
8.54
0.084
21.5
3.17
10.03
0.008
41.75
9.25
19.33
n 1.6
r 2.55
1 max 51.3
the performance of aperture antennas such as lenses and reflectors. With a perfect lens and a point source at its focus, the
phase error should not exist. However, there are fabrication
tolerances, and misalignments can occur that will contribute
to aberrations. Even without such imperfections, lens antennas can suffer from aberrations. Practical lens feeds are horn
antennas and small arrays. Both have finite sizes and deviate
from the point source (2). This means that part of the feed
aperture falls outside the focal point, and rays emanating
from them do not satisfy the optical relationships. Thus, on
the lens aperture the phase distribution is not uniform. Similar situations also occur when the feed is moved off axis laterally to scan the beam. Again, aperture phase error occurs due
to the path length differences. A somewhat different situation
arises when the feed is moved axially, front or back. In this
case the phase error is symmetric, because all the rays leaving the source with equal angles travel equal distances and
arrive at the aperture at an equal radial distance from the
axisthat is, on a circular ring. However, the length of the
ray increases, or decreases, with radial distance on the aperture. The phase error is, therefore, quadratic on the aperture
and reduces the aperture efficiency, while raising the sidelobes.
The general aberration (i.e., the lens aperture phase error)
can depend implicitly on both feed and lens coordinates and
can be difficult to comprehend. However, like all other phaseerror-related problems, it can also be represented as the path
length difference with a reference ray. For rotationally symmetric rays, the natural reference is the axial ray. The path
length difference can then be obtained by a Taylor-type
expansion of the general ray length in terms of the axial one.
For small aberrations the first few terms in the expansion
will be sufficient to describe the length accurately. In terms of
the aperture polar coordinates and the expansion becomes
ing a Fourier transformation or diffraction integral. For onedimensional errors (i.e., x and o) the effect can be
understood easily and has been investigated by Silver (1). The
first term is linear, and in a Fourier integral it shifts the
transform variable. It thus causes a tilt of the beam, but the
gain remains the same. Using Silvers notation, if f(x) is the
aperture distribution and g(u) is the far field (i.e., its Fourier
transform with a linear phase error), one finds with no phase
error
g0 (u) =
a
2
1
1
(24)
g(u) =
a
2
1
1
(25)
a
a 1
f (x) exp[ j(ux x2 )] dx
2 1
a
g(u) =
(26)
(27)
DIELECTRIC-LOADED ANTENNAS
degradation than others, it is desirable to eliminate it, especially that it manifests mostly in beam scanning. Feed lateral
displacements to scan the beam can readily cause coma lobes.
Fortunately, a number of lens surface modifications have
been found to reduce the effects of this error (3).
Gain decibels
= /2
10
285
= /4
ZONED LENSES
20
No phase error
30
3
u = ( a/ ) sin
a 1
f (x) exp[ j(ux x3 )] dx
2 1
a [g (u) + g (u)]
=
0
2 0
g(u) =
(28)
So far, the equations used for lens design equalized the ray
path lengths. The frequency of operation, or its wavelength,
did not enter into the equations. Thus, in principle, these
lenses should function at all frequencies. However, the directivity of a lens depends on its aperture size D, and these
lenses are often used for high gain applications. This results
in large lens sizes in wavelength; and at microwave frequencies, in large physical sizes, both the aperture diameter D and
thickness T. It can, therefore, become excessively heavy and
difficult to use. However, the thickness of the lens can be several wavelengths, and thus it can be reduced along the ray
path in multiple wavelengths without altering the relative
phase change. The process starts at the edge where the thickness is zero. Moving down toward the axis, the thickness increases progressively until it becomes one wavelength. This
thickness can be made zero without altering the phase. The
process can be continued K times until one arrives at the lens
axis. In practice, one must maintain a small thickness tm to
provide adequate mechanical strength, the value of which will
depend on the lens size, the material strength, and application type.
With zoned lenses, and neglecting tm because the thickness
does not exceed one electrical wavelength, its thickness is limited to /(n 1). Including the minimum thickness tm, the
total thickness is limited to tm l/(n 1) regardless of the
number of zones. The path lengths in wavelength, however,
are not equal. With K zones, the ray path at the edge will be
longer by a length equal to (K 1). This causes the frequency dependence of lens operation, limiting its bandwidths.
Enforcing the commonly used Silvers criterion for this aperture phase error (1) (i.e., less than 0.125), the useful bandwidth of a lens with K zones can be calculated from (1)
Bandwidth
=
25
%
K 1
(29)
0
No phase error
Gain decibels
= /4
10
= /2
20
30
3
u = ( a/ ) sin
286
DIELECTRIC-LOADED ANTENNAS
Shadow rings
R2
R1
S
Focus
K=1 K=2 K= 3
F
tm
ing will be eliminated, but phase errors still occur at the transition lines due to diffraction effects.
(1 n)F
1 n cos 1
(30)
CONSTRAINED LENSES
The function of a lens is to modify the phase front of an incident wave, say from spherical to planar. In practice, this may
be accomplished by means other than the dielectric lenses. In
the most general case, the lens surfaces consist of a plurality
of receiving and radiating elements, interconnected by processing elements. The received signals of one surface are modified in amplitude and phase and are reradiated from the elements of the next surface. In passive designs, the
interconnection is due to transmission lines, such as parallel
plates, waveguides, and even coaxial lines. The design process
is similar to that of the dielectric lenses and is governed by
the path-length equation. Snells law, however, is not satisfied
at all surfaces, and the problem of surface reflection and
transmission must be solved through the use of the wave
equation. Nevertheless, lenses can be designed with surfaces
similar to that of dielectric lenses, but with inverted curvature (3).
The simplest case uses parallel plates, with spacing a, between one and one-half wavelength. When the electric field is
parallel to the plates, a non-TEM waveguide mode is excited
and has a wavelength p given in terms of the free space
wavelength by
p =
1
2a
2 1/2
(31)
<1
(32)
= 1
n=
p
2a
In cylindrical lenses, when the plates and electric field are
normal to the cylinder axis, Snells law of refraction governs
DIELECTRIC-LOADED ANTENNAS
the transition between the lens and outside media. But, when
they are parallel to the cylinder axis, the incident rays are
constrained to pass between the plates and Snells law is not
satisfied (1).
An example of the rotationally symmetric constrained lens
is the planarelliptic surface lens of Eq. (30). It is usually
zoned to reduce its size and weight (4). Other useful transmission media are the rectangular and square waveguides, operating in TE10 or TE01 modes. The waveguide dimensions
must be such that only these modes can propagate and
higher-order modes are suppressed. The square waveguide
can be used for circularly polarized applications; otherwise, it
must be avoided to reduce cross-polarization.
INHOMOGENEOUS LENSES
In the lenses so far studied, the refractive index n was constant and the shape was profiled to satisfy the ray path condition. On the other hand, if the lens shape is kept fixed, then
another parameter, such as the refractive index, must be allowed to change to help in collimating the beam. This is
achieved in a family of lenses, with the most important ones
being spherical in shape, such as Luneberg lens, Maxells fisheye, and Eaton lenses. Their spherical shape provides a perfect three-dimensional symmetry, useful in applications, such
as wide-angle scanning. They also have only a radial inhomogeneity, making them both physically and electrically symmetric.
Luneberg Lenses
The Luneberg lens refers to a family of lenses with two axial
foci. They can be both outside the lens or one inside and the
other outside. The most useful case, however, is the lens with
one focus on its surface, while the second one is at infinity.
That is, an axial point on the lens surface is focused to an
axial point at infinity, on the opposite side of the lens. The
refractive index of this lens is given by
r 2 1/2
n(r) = 2
a
287
Limits of solid
angular coverage
determined by
boundaries of cap
reflectors
Spherical cap
reflector
Lens
(33)
Lens
Feed antenna
Figure 9. Typical ray paths in a Luneberg lens.
288
DIELECTRIC-LOADED ANTENNAS
60
2.4
r = 1.5
50
2.2
40
2.0
Normalized phase (deg)
Dielectric constant
1.8
1.6
d
a
1.4
c
1.2
r = 2.0
30
20
r = 2.5
10
r = 3.0
0
r = 3.5
10
r = 4.0
20
30
0.2
0.6
0.8
0.4
Radial distance (r/a)
1.0
40
60
the lens diameter in wavelength is large. At microwave frequencies, the wavelength is large and the lens diameter in
wavelength may not be large. Its performance (i.e., directivity) and sidelobe levels deteriorate rapidly. In such cases, the
refractive index profile can be modified to improve its performance. This can be done by determining the excitation efficiencies of various spherical modes and calculating its far
field and directivity (5). The new dielectric permittivity profile
is defined as
r = n2 = 2B A2
r 2
(34)
r = 6.0
50
0.8
AB1
Diameter
in
2
4
6
8
10
A1
A2 B
B Value
Gain
(dB)
Beam
Width
(degrees)
Gain
(dB)
Beam
Width
(degrees)
Gain
(dB)
Beam
Width
(degrees)
1.4
1.16
1.1
1.075
1.04
14.79
20.761
24.34
26.90
28.78
30.2
15.1
9.8
7.3
5.8
14.41
16.05
16.9
17.1
16.35
17.56
22.70
25.75
27.98
29.35
23.5
13.0
9.0
6.7
5.5
17.15
16.9
16.97
17.01
15.97
16.85
22.0
25.17
27.56
29.26
24.0
13.25
9.1
7.0
5.6
14.79
16.1
16.4
16.6
16.81
yyy
;;;
;;;
yyy
;;;
yyy
;;;
yyy
;;;
yyy
DIELECTRIC-LOADED ANTENNAS
289
(a)
(b)
ture diameter, and both are dependent on the dielectric permittivity. The air gap size generally increases with the horn
diameter, and for a given diameter there is a minimum relative permittivity of dielectric to support the hybrid mode to
minimize the cross-polarization. Both flare angle and the
throat region have similar influences. Large flare angles and
an asymmetric throat region design excite higher-order
modes and thus increase cross-polarization.
A variation of the conical dielectric loaded horn is shown
in Fig. 15. Its wall is profiled. A large flare angle near its
throat reduces its axial length and results in a compact horn.
Then, its small flare angle near the aperture improves its
cross-polarization. The profile is described by the following
equation:
r(z) = rth + 3r 1
2z
3L
z 2
L
(35)
r = rap rth
yyy
;;;
;;;
yyy
;;;
yyy
where rap and rth are the horn radii at its aperture and throat.
A profile horn of this type was designed and optimized. Its
290
DIELECTRIC-LOADED ANTENNAS
Linear
Horn
Profiled
Horn
14.8
26.9
22.1
61.8
32.2
1.04
13.7
24.8
22.5
68.1
36.0
1.03
Rth 1.14 cm
rup 27.7 cm
L 30.9 cm
r 1.13
Air gap 1.2 cm
max
D
T
(a)
max
F
(b)
r
Peak
Cross-Polar
(dB)
3 dB
Beamwidth
(deg)
Efficiency
%
30.9
27.5
24.0
15.0
36.0
36.8
31.6
27.6
13.7
13.8
14.0
16.1
68.1
64.4
57.1
32.0
T
(c)
are shown in Fig. 17. It shows that for r around 1.22 the
aperture amplitude distribution is nearly uniform.
DIELECTRIC LOADED WAVEGUIDES
Waveguides have small aperture size and are not as efficient
radiators as horns. Part of the energy leaks out and induces
current on the outside wall, which radiates side and backward, causing large back lobes. The wave impedances of
waveguide modes are also different from the free-space intrinsic impedance and strong reflections can occur on the aperture, causing poor input impedance match. These problems
can be partly overcome by flaring the waveguide at its aperture. However, a similar and even better performance can be
obtained by loading the waveguide by a short section of a dielectric. The constant, size, and shape of a dielectric provide
several parameters that can be used to shape the radiation
patterns and tailor them to desired specifications. Table 5
shows the results for three different end loadings, along with
the type of performance variations one could achieve (2). Two
other examples are shown in Figs. 18 and 19, with combinations of dielectric and cavity loadings (2). In Fig. 18, the end
geometry is optimized for nearly perfect pattern symmetry,
with negligible cross-polarization. Figure 20 shows its copolar
and cross-polar radiation patterns. In Fig. 19, again the com-
yyy
;;;
;;;
yyy
DIELECTRIC-LOADED ANTENNAS
18
291
37.5 mm
1.1
r = 2.6
12
26 mm
61 mm
32 mm
1.2
6
dB
20
10
30
(deg)
6
1.23
Horn alone
1.5
14
a ground plane by a dielectric substrate. They are low profile and increasingly popular antennas for practically any
type of applications. Their radiation patterns, however, are
asymmetric with unequal E- and H-plane patterns. But,
with careful optimization, the pattern symmetry can be
achieved to minimize cross-polarization. Figure 21 shows a
case of stacked patches with a side choke for equalizing
the principal plane pattern, low back radiation, and crosspolarization. Similar performance can also be obtained using a dielectric resonator in lieu of a microstrip patch. The
dimensions of the dielectric resonator are related to the
wavelength by
d 2 1/2
1.841
16 +
d=
4n
1.841h
(36)
Geometry
Half-Beamwidths
Peak
Cross-Polarization
0 90 (dB)
Gain
(dB)
E Plane
H Plane
E Plane
H Plane
33.95
8.28
36.82
36.18
71.47
72.51
24.74
8.11
37.21
38.32
73.42
71.35
24.43
13.47
19.43
20.25
33.13
35.17
3 dB
10 dB
0.519
60
0.1
0.6
b
60
0.6
c
0.619
D = 0.6, r = 2.5
292
DIELECTRIC-LOADED ANTENNAS
E plane
H plane
Cross-plane
10
16
24
20
32
30
40
36
72
108
144
180
(deg)
Figure 19. Geometry and radiation pattern of a shaped dielectric
and cavity loaded waveguide feed.
40
60
120
180
(deg )
of 2a, insulated by a cylindrical dielectric region of diameter 2b and propagation constant k1, located in an infinite
exterior region of k2. With a thin-wire approximation, the
dipole current can be represented by a sinusoidal distribution of the form (10). The time factor is assumed as
exp( jt):
I(z) =
jV sin kL (h |z|)
2Zca cos kL h
(37)
0
Relative power one way (dB)
INSULATED ANTENNAS
16
24
32
40
180
135
90
45
0
(deg)
45
90
135
180
DIELECTRIC-LOADED ANTENNAS
293
2b
3t
2a
K1 K2
d
h
+
v
D
where
1/2
H0(2) (k2 b)
k L = k 1 1 +
b
k2 bH1(2) (k2 b)ln
a
Zca =
(38)
and so on. Thus, the insulation parameters should be selected
appropriately to minimize the dependence of k1 and k2.
b
1 k L
ln
2k1
a
(39)
0
k1
(40)
1 =
k1 = [1
1 ]
1/2
(41)
(2)
And H(2)
0 and H1 are Hankel functions of zero and first order.
Note that with a perfect insulation dielectric, k1 is real; but
k2 is complex due to the presence of Hankel functions in Eq.
(38). It reduces to k1 when b, the radius of the insulation,
becomes infinitely large. In view of Eq. (38) the dipole current
distribution, input impedance, radiation resistance, and resonance frequency can depend strongly on the radius b, propagation constant k1, and the propagation constant of the exterior region, k2. The latter may not be fully known, or constant,
during the application due to variations in moisture content,
0
E plane
H plane
Cross-plane
Relative power (dB)
10
20
eff =
30
40
180
90
0
(deg)
90
180
r + 1
+
2
1
2
1+
12h
w
1/2
(42)
294
DIELECTRIC MEASUREMENT
reduced by introducing a superstrate layer over the microstrip antenna to control the relative permittivity variations.
Invasive-type radiators can produce more uniform and
controllable heating patterns, but they require implantation
in the tissue. The most convenient types are the insulated
needle radiator, basically the end of the coaxial line. However, this type of antenna can generate strong currents on
the outer coaxial conductor and cause tissue heating behind
the antenna. An improvement can be obtained by introduction of a quarter-wavelength choke over the coaxial conductor to form a sleeve antenna. Their analysis and sensitivity
study can be carried out similar to the insulated dipole
antennas. Figure 25 shows the geometry of needle and
sleeve antennas.
BIBLIOGRAPHY
1. S. Silver, Microwave Antenna Theory and Design, London: P.
Peregrinus, 1984.
2. A. D. Oliver et al., Microwave Horns and Feeds, London: P. Peregrinus, 1994.
3. Y. T. Lo and S. W. Lee, Antenna Handbook, Theory Applications
and Design, New York: Van Nostrand Reinhold, 1988, chap.
16.
4. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, 2nd
ed., New York: McGraw-Hill, 1984.
5. M. Barakat and L. Shafai, Studies on certain modified Luneberg
lenses, IEE Proc., 130, part H (5): 363368, 1983.
6. G. Bekefi and G. W. Farnell, A homogeneous dielectric sphere as
a microwave lens, Can. J. Phys., 34: 790803, 1956.
7. V. B. Mason, The electromagnetic radiation from simple sources
in the presence of a homogeneous dielectric sphere, PhD dissertation, Univ. of Michigan, 1972.
8. P. J. B. Clarricoats, A. D. Oliver, and M. Rizk, A dielectric loaded
conical feed with low cross-polar radiation, Proc. URSI Symp. EM
Theory, Spain, 1983, pp. 351354.
9. E. Lier, A dielectric hybrid mode antenna feed, a simple alternative to the corrugated horn, IEEE Trans. Antennas Propag., AP34: 2129, 1986.
10. R. W. P. King et al., The insulated monopole: Admittance and
junction affect, IEEE Trans. Antennas Propag., AP-23: 172
177, 1975.
11. I. J. Bahl and S. S. Stuchly, Analysis of a microstrip covered with
a lossy dielectric, IEEE Trans. Microw. Theory Tech., MTT-28:
104109, 1980.
L. SHAFAI
University of Manitoba
DIPOLE ANTENNAS
INTRODUCTION
1
jko
1
ILsin(
+ 2 )e jko r
4
r
r
jo IL
jko
1 jko r
E=
cos( 2 + 3 )e
r
2ko
r
r
2
jo IL
k
1
jko
sin( o + 2 + 3 )e jko r
4ko
r
r
r
H=
(1)
(2)
j
ILsine jko r
4r
(3)
E=
jo
ILsine jko r
4r
(4)
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Dipole Antennas
Figure 3. (a) Current distributions. (From Reference 3.) (b) Radiation patterns for a 1.25 dipole antenna.
I(0)
sink(h |z|) f or h < z < h
sinkh
(5)
jI(0)
F ()e j(tkro )
2ro sin(kh)
(6)
H=
jI(0)
F ()e j(tkro )
2ro sin(kh)
(7)
and
cos(khcos()) coskh
sin
(8)
E
E
I 2 (0)
F 2 ()
r =
2
2
8 r2 sin2 (kh)
(9)
Figure 4. Variation of feed-point Ra and reactance Xa for an endfed monopole above ground as a function of height h/. Values for
dipoles are double that of monopoles. (From Reference 4, p 543.)
30 W F 2 ()
ro2 Ra sin2 kh
(10)
Pmax
=
Po
F 2 ()max
1
2
0
F 2 ()sind
120 F 2 ()|max
Ra sin2 kh
(11)
Dipole Antennas
2
4Za
(12)
Dipole Antennas
Table 1. Wire Lengths Required to Produce a Resonant Half-Wave Dipole for a Wire Diameter of 2a and a Length L
Length-to-diameter ratio, L / (2a)
Resonant length
2
5
9
0.49
0.475
0.455
Very thin
Thin
Thick
5000
50
10
drical dipoles at the ends and driving the entire structure by a transmission line (often a two-wire transmission
line) at the center of one arm as shown. The feed-point
impedance of a folded dipole of two identical-diameter arms
is four times as large as for an unfolded dipole of the same
length, which can actually be advantageous, because the
feed-point resistance may now be comparable with the
characteristic impedance Zo of the transmission or feed
line. The reactance of the antenna may easily be compensated by using a lumped element with a reactance that
is negative of the reactance at the terminals of the folded
dipole antenna or else by using a foreshortened antenna
length to resonant length arms so that Xa = 0 (see Table
1).
Shunt-Fed Dipoles. Matching networks of reactive elements are generally required to match the feed-point
impedance (Ra + j Xa ) of center-fed dipoles to transmission
lines. Typically these lines have characteristic impedance
on the order of 300 to 600 , and a thin half wave dipole
has impedance Z = 73 + j 42 . To alleviate the need for
matching networks, the dipoles are at times shunt-fed at
symmetric locations off the center point as shown in Fig.
9. This procedure using either the delta match (Fig. 9a) or
the T-match (Fig. 9b) is often used for half-wave dipoles
(2h = /2) with A and B dimensions that are typically on
the order of 0.10 to 0.15 .
Figure 10. Sleeve dipole antenna: (a) physical model and (b)
equivalent electrical model.
Sleeve Dipole. The sleeve dipole antenna and its equivalent electrical model are shown in Fig. 10. In practice,
this antenna is built from a coaxial line with the outside
conductor and insulation stripped away from the center
conductor, which is left protruding. The outer conductor is
connected to the ground plane, and the image produced by
the ground plane creates an equivalent sleeve dipole antenna. These dipoles are useful, because they have a broad
band VSWR over nearly an octave of bandwidth.
Dipole Antenna Arrays
Dipole antennas and arrays of dipole antennas are used
for short wave (3 MHz to 30 MHz) and for VHF and UHF
(30 MHz to 900 MHz) radio and TV broadcasting. If directional communication is desired such as for short-wave
radio transmission via the ionosphere, a phased array of
horizontal dipoles may be used mounted above a ground
plane. The spacing is chosen to send the major lobe of radiation toward the sky at a suitable angle to reect off the
ionosphere and to provide broadcast coverage over the desired service area.
For VHF and UHF radio and TV broadcasting over a
360 azimuthal angle, colinearly mounted vertical dipoles
that are excited in-phase with each other are often used.
Two examples of this are shown in Fig. 11a and b. An example variation of this is a three- to eight-bay turnstile
Dipole Antennas
Figure 13. Log-periodic dipole array: (a) geometry of a logperiodic array showing how the phase-reversal feed system for
this antenna is constructed. (From Reference 8.) (b) Equivalent
antenna model of the log-periodic array. Mc Graw-Hill 1993.
Rn+1
dn+1
=
Rn
dn
(13)
dn
(1 )
cot
=
2Ln
4
(14)
Dipole Antennas
plane or wire mesh. This array has several desirable features, including high gain, broad bandwidth, independent
control of horizontal and vertical radiation patterns, ease of
matching (low VSWR), and the ability to efciently broadcast efciently. Using a phased-feed system, this array allows beam steering of the radiation pattern in both the
azimuthal and the elevation planes, providing a very high
degree of exibility.
YagiUda Dipole Array. YagiUda arrays are commonly
used as general-purpose antennas from 3 MHz to 3000
MHz, in particular, as home TV antennas. They are inexpensive, have reasonable bandwidth, and have gains up
to 17 dBi or more if multiple arrays are used. They have
unidirectional beams with moderate side lobes (8).
A typical YagiUda array is shown in Fig. 16. This array
is a simple end-re array of dipole antennas where only one
element is driven and the rest are parasitic. The parasitic
elements operate as either reectors or directors. In general (8), the longest antenna, which is about /2 in length, is
the main reector and is generally spaced /4 in back of the
driven dipole. The feed element is commonly a folded dipole
antenna 0.45 to 0.49 long. Adding directors, which are
generally 0.4 to 0.45 long, to the front of the driven
element increases the gain of the array. The directors are
not always of the same length, diameter, or spacing. Common arrays have 6 to 12 directors and at most 2 reectors.
Additional improvements in gain by adding more elements
are limited; however, arrays have been designed with 30 to
40 elements (3). A gain (relative to isotropic) of 5 to 9 per
wavelength of array length is typical for YagiUda arrays,
for an overall gain of 50 to 54 (14.8 to 17.3 dB).
The YagiUda array is characterized by a main lobe
of radiation in the direction of the director elements and
small side lobes. The beamwidth is small, generally 3060
(3). Typical E and H plane patterns of a YagiUda array are
shown in Fig. 17. Typically, the performance of a Yagi-Uda
array is computed using numerical techniques (10). For
Dipole Antennas
Figure 17. Typical E and H plane patterns of a YagiUda array. Total number of elements = 27, number of directors = 25, number of
reectors = 1, number of driven elements = 1, total length of reector = 0.5 , total length of feeder = 0.47 , total length of each director =
0.406 , spacing between reector and feeder = 0.125 , spacing between adjacent directors = 0.34 , and radius of wires = 0.003 . (From
Reference 9.) IEEE 1969.
resistively loaded bow-tie antennas in the shape of a Maltese cross shown in Reference 24 has also been proposed to
pick up the cross-polarized reections (18). Cross-polarized
reections from simple tumor models were also examined
in References 24 and 25.
The antenna shown in Figure 21 (24) consists of two
cross-polarized bow-tie antenna elements, an octagonal
cavity behind the bow-tie elements, and a metal ange attached to the cavity. The broadband bow-ties have are
angles of 45 . They are 1.67 cm long, which is a halfwavelength at 3 GHz in fat (similar to breast). The octagonal cavity blocks waves radiated away from the breast.
The cavity is approximated as a circular waveguide lled
with fat material for matching and size reduction. The rst
cutoff frequency is set to be 2 GHz for 24 GHz operation.
The cavity length is a quarter-wavelength, which is 11 mm
at 3 GHz. The ange consists of an inner and outer component and is designed to block unwanted waves such as
surface waves. The antenna performance does not change
signicantly when the ange size is varied between 10 and
6.25 cm; therefore, the width of the outer ange is set to be
6.25 cm. The inner ange is designed to prevent possible
electric eld overshoot at the inner corners of the opening
of the octagonal cavity or at the ends of the bow-tie elements. A slotline bow-tie antenna has also been proposed
in Reference 26.
Vector Antennas. Loop antennas have received considerable attention lately in the development of compact wireless communication systems when integrated with other
antennas such as dipoles. This use of loop antennas leads to
the possible development of compact Multiple Input Multiple Output (MIMO) systems, which can nd applications
not only in communication systems, but also in the direction of arrival estimation, sensor networks, and imaging.
Although, traditionally, the increase in channel capacity in
Dipole Antennas
Figure 19. Geometry and dimensions of the proposed dual ISMband antenna. (From Reference (13), IEEE 2005.)
a communication system has been achieved through spatial array of antennas, similar increases in channel capacity have been obtained through the use of vector antennas consisting of colocated loops and dipoles, which can
respond to more than one component of the electromagnetic eld. One example of such antennas is given in Reference 27, where three- and four-element vector antennas,
consisting of one loop and two or three dipoles, respectively,
are employed in a MIMO system. Several research efforts
that address the characterization of systems of colocated
loops and dipoles can be found in the literature (2732).
Dipole Antennas
10
Dipole Antennas
33. Kandonian, A. G. Three New Antenna Types and Their Applications. Waves Electrons 1946.
Reading List
ITT Handbook. Reference Data for Radio Engineers, 5th ed.;
Howard W. Sams: Indianapolis, In, 1973.
Gandhi, O. P.; Chen, J. Y. Electromagnetic Absorption in Human Head from Experimental 6-GHz Hand-Held Transceivers.
IEEE Trans. Electromagn. Compat. 1995, 37, pp 547558.
Tai, C. T. On the Theory of Biconical Antennas. J. Appl. Phys. 1948,
pp 11551160.
King, R. W. P. Theory of Linear Antennas; Harvard University
Press: Cambridge, MA, 1956.
Ma, M. T.; Spies, K. P. A Simplied Formulation Computation for
Conical Monopole Antennas. U.S. Dept. of Commerce Report OT
74-53, 1974.
Hicks, D. E. CB Radio Antennas; Howard W. Sams: Indianapolis,
IN, 1967.
The A.R.R.L. Antenna Book; The American Radio Relay League:
West Hartford, CT, 1956.
Johnson, R. C.; Jasik, H. Antenna Engineering Handbook, 2nd ed.;
McGraw-Hill: New York, 1984.
Lazzi, G., et al. Comparison of FDTD-Computed and Radiation
Patterns of Commercial Mobile Telephones in the Human
Head. IEEE Trans. Antennas Propagat. 1998, 46.
Luebbers, R., et al. FDTD Calculation of Radiation Patterns,
Impedance and Gain for a Monopole Antenna on a Conducting Box. IEEE Trans. Antennas Propagat. 1992, 40, pp 1577
1583.
Jensen, M. A.; Rahmat-Samii, Y. EM Interaction of Handset Antennas and a Human in Personal Communications. Proc. IEEE
1995, 83, pp 717.
Okoniewski, M.; Stuchly, M. A. A Study of the Handset Antenna
and Human Body Interaction. IEEE Trans. Microw. Theory
Tech. 1996, 44, pp 18551864.
Lazzi, G. et al. Comparison of FDTD-Computed and Radiation Patterns of Commercial Mobile Telephones in the Human Head.
IEEE Trans. Antennas Propagat. 1998, 46.
Gandhi, O. P.; Lazzi, G.; Furse, C. M. Electromagnetic Absorption
in the Human Head and Neck for Mobile Telephones at 835
and 1900 MHz. IEEE Trans. Microw. Theory Tech. 1996, 44, pp
18841897.
Special Issue of IEEE Trans. Microw. Theory Tech. 1986, MTT-34.
Durney, C. H.; Iskander, M. F. Antenna Handbook, Lo, Y. T., Lee, S.
W. Eds.; 1993.
Sneed, P. K.; Phillips, T. L. Combining Hyperthermia and Radiation: How Benecial? Oncology, 1991, 5, pp 99108.
Vernon, C. C.; Hand, J. W.; Field, S. B., et al. Radiotherapy With or
Without Hyperthermia in the Treatment of Supercial Localized Breast Cancer: Results from Five Randomized Controlled
Trials. Int. J. Radiat. Oncol. Biol. Phys. 1996, 35, pp 731744.
Turner, P. F. Interstitial Equal-Phased Arrays for EM Hyperthermia. IEEE Trans. Microw. Theory Tech. 1986, 34, pp 572
578.
Furse, C. M.; Iskander, M. F. Three-Dimensional Electromagnetic
Power Deposition in Tumors using Interstitial Antenna Arrays.
IEEE Trans. Biomed. Eng. 1989, 36, pp 977986.
Nevels, R. D.; Arndt, G. D.; Raffoul, G. W.; Carl, J. R.; Pacico, A.
Microwave Catheter Design. IEEE Trans. Biomed. Eng. 1998,
45, pp 885890.
Manry, C.; Broschat, S. L.; Chou, C.-K.; McDougall, J. A. An Eccentrically Coated Asymmetric Antenna Applicator for Intracavity
Hyperthermia Treatment of Cancer. IEEE Trans. Biomed. Eng.
1992, 39, pp 935942.
Fear, E. C.; Stuchly, M. A. Microwave Breast Tumor Detection: Antenna Design and Characterization. IEEE Antennas Propagat.
Symp. Dig. 2000, 2, pp 10761079.
Shannon, C. J.; Fear, E. C.; Okoniewski, M. Dielectric-Filled Slotline Bowtie Antenna for Breast Cancer Detection. Electron.
Lett. 2001, 41.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer Detection: A Study of Immersion Liquid. Electron. Lett.
2005, 41, pp 113115.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer Detection: Preliminary Experimental Results. Proc.
IEEE MTT-S Int. Microw. Symp. Dig.; Long Beach, CA, June
2005.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer DetectionExperimental Investigation of Simple Tumor Models. IEEE Trans. Microw. Theory Tech. 2005, 53, pp
33123319.
Wu, T.; King, R. The Cylindrical Antenna with Nonreecting Resistive Loading. IEEE Trans. Antennas Propagat. 1965, AP-13,
pp 369373.
Wu, T.; King, R. Corrections to The Cylindrical Antenna with Nonreecting Resistive Loading. IEEE Trans. Antennas Propagat.
1965, AP-13, p 998.
Labonte, S., et al. Monopole Antennas for Microwave Catheter
Absorption. IEEE Trans. Microw. Theory Tech. 1996, 44, pp
18321840.
Wolff, E. A.. Antenna Analysis; Wiley: New York, 1966.
Werner, D. H. An Exact Integration Procedure for Vector Potentials of Thin Circular Loop Antennas. IEEE Trans. Antennas
Propagat. 1996, 44, pp 157165.
Smith, G. S. Loop Antennas. In Antenna Engineering Handbook;
McGraw-Hill: New York, 1984.
Storer, J. E. Impedance of Thin-Wire Loop Antennas. AIEE Trans.
1956, 75.
Greatbatch, W.; Holmes, C. F. History of Implantable Devices.
IEEE Eng. Med. Biol. 1991, pp 3842.
Woolons, D. J. To Beat or Not to Beat: The History and Development of Heart Pacemakers. IEE J. Eng. Sci. Educ. 1995, 4, pp
259268.
Allan, R. Medtronic Sets the Pace with Implantable Electronics.
Electron. Design 2003, 51, pp 5256.
Spelman, F. A. The Past, Present, and Future of Cochlear Prostheses. IEEE Eng. Med. Biol. 1999, pp 2733.
Rauschecker, J. P.; Shannon, R. V. Sending Sound to the Brain.
Science 2002, 295, pp 10251029.
Weiland, J. D.; Humayun, M. S. A Biomimetic Retinal Stimulating
Array. IEEE Eng. Med. Biol. Mag. 2005, 24, pp 1421.
Walter, P.; Kisvarday, Z. F.; Gortz, M.; Alteheld, N.; Rossler, G.;
Stieglitz, T.; Eysel, U. T. Cortical Activation via an Implanted
Wireless Retinal Prosthesis. Invest. Ophthalmol. Visual Sci.
2005, 46, pp 17801785.
Margalit, E.; Maia, M.; Weiland, J. D.; Greenberg, R. J.; Fujii, G.
Y.; Torres, G.; Piyathaisere, D. V.; OHearn, T. M.; Liu, W.; Lazzi,
G.; Dagnelie, G.; Scribner, D. A.; de Juan, E.; Humayun, M. S.
Retinal Prosthesis for the Blind. Surv. Ophtalmol. 2002, 47, pp
335356.
Dipole Antennas
Normann, R. A.; Maynard, E. M.; Guilloty, K. S.; Warren, D. J.
Cortical Implants for the Blind. IEEE Spectrum 1996, pp 54
59.
Zrenner, E. Will Retinal Implants Restore Vision? Science 2002,
295, pp 10221025.
Nicolelis, M. A. L. Brain-Machine Interfaces to Restore Function
and Probe Neural Circuits. Nature Rev. Neurosci. 2003, 4, pp
417422.
Chapin, J. K.; Moxon, K. A. Eds. Neural Prostheses for Restoration
of Sensory and Motor Function. CRC Press: Boca Raton, FL,
2000.
Ghovanloo, M.; Lazzi, G. Transcutaneous Magnetic Coupling of
Power and Data (invited paper). In Wiley Encyclopedia of
Biomedical Engineering; Akay, M. Editor, Wiley (to appear,
2006).
CYNTHIA M. FURSE
OM P. GANDHI
GIANLUCA LAZZI
University of Utah
North Carolina State University
11
DIRECTIVE ANTENNAS
589
Direction of wave
DIRECTIVE ANTENNAS
Radio antennas that are directive have many advantages
over stationary antennas. They can observe in one direction
while ignoring much larger signals coming in from the
sides. In addition, as transmitting antennas, they can direct
a small amount of radiation directly to the proper receiver.
Conventional directive antennas use large mirrors for short
waves, such as are used in radar. For longer-wave antennas, they use reflector and director elements aligned with
the antenna. The basic problem with such arrays is that
they are mechanically steered, which is slow and difficult
to do with large antennas, especially in the presence of
wind, ice, or snow.
One way of having a large steerable antenna with no moving parts is to use a phased array. A phased array is based
on Huygens principle, which states that a wave front can be
determined at a point in time by constructing a surface tangent to a collection of secondary waves. Thus, if one has a
large number of small antennas located on a line, and if the
antennas are excited in phase, the resultant wave front is
normal to the line. If each small antenna is excited with a
small phase shift relative to the next antenna, the wave propagates at an angle to the line. Conversely, if the array is used
for reception, the small antenna signals, combined with the
individual phase shifts, allow the array to scan in azimuth.
The principle can be improved by using a two-dimensional
array located on a plane to scan in altitude as well as azimuth. In fact, by properly adjusting the phase shifts, such an
IGOR ALEXEFF
MARK RADER
WENG LOCK KANG
The University of Tennessee
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
752
DIVERSITY RECEPTION
DIVERSITY RECEPTION
Diversity reception is a radio communication technique that
improves system performance during periods of adverse propagation conditions by providing more than one transmission
channel (or branch) to deliver the signal intelligence to a specified destination. Generally, the goal is to increase the transmission link availability sufficiently to meet prescribed system performance criteria and provide acceptable service. To
take advantage of the multiple channels and increase the link
performance, the capability must be provided either to select
among the available signals, or else to combine the signals.
(An approximate analogy is found in telephony, where many
paths are generally available to the Public Switched Telephone Network to complete a call from one telephone to another, greatly increasing the probability of completing any
given call.)
Usually, the objective in diversity reception is to reduce
performance degradations caused by signal fading, such as
multipath fading in mobile and terrestrial point-to-point systems, or signal attenuation caused by rainfall on the propagation path (rain attenuation) in earth-space (satellite) systems.
Significant performance improvements can also be achieved
with respect to other path impairments, such as unwanted
signal depolarization (as encountered in dual-polarization frequency-reuse communication systems), angle-of-arrival variations, and cochannel interference. In many scenarios, link
performance can of course be enhanced by simply increasing
the transmitter power (assuming this approach is cost-effective), but this option is often precluded by regulations established to limit intersystem interference.
The propagation medium is presumed to cause occasional
degradations in a single transmission channel that are sufficiently severe to justify the expense and complexity of implementing diversity reception. A sufficient understanding of the
propagation environment is essential for designing effective
diversity reception systems. For example, when multiple
earth terminals are installed in earth-space telecommunication systems to reduce rain attenuation outages that would
be experienced on a single path, the minimum site separation
for the diversity terminals is dictated mainly by characteristics of the rain environment, although performance elements
of the earth terminals, such as antenna gain and link fade
margin, are also quite important.
For diversity reception to be effective, impairments on the
separate channels are preferably independent, or at least sufficiently decorrelated that simultaneous severe signal degradations are rare. If the time-varying propagation effects on
the individual channels are highly-correlated, the probability
of simultaneous signal impairments is large, and the benefits
offered by diversity reception will be small. (Anticorrelation
of impairments is even more advantageous than zero correla-
tion, but this condition is not generally attained.) If the propagation medium is spatially uniform or is not time-varying,
there is little reason to use diversity. Diversity reception is
not generally intended to counter slowly-varying macroscopic
(bulk) changes in the propagation environment, as such
changes tend to affect all available channels more-or-less
equally. A general condition to be met in diversity systems is
that the individual diversity signals should have similar
mean received power levels (within 10 dB or so). Otherwise,
the link performance is dominated by the strong signal(s),
with little gain derived from the other channels.
Providing redundant transmission channels to deliver
identical information can be expensive, and inevitably increases the equipment complexity. For example, diversity installations that employ spatially separated antennas must be
connected by a communication link, such as a microwave link
or optical cable, to allow combining of, or switching among,
the signals from the diversity branches. In some installations,
the diversity terminals may be separated by many kilometers.
Furthermore, conditioning of the multiple signals is typically
required to support selection or combining of the signals without losing information (hitless switching), and a decision
criterion or algorithm must be devised to control any diversity
operation. There must usually be an expectation of substantial performance benefits to justify implementation of diversity reception.
A well-designed diversity reception system can yield impressive enhancements in system performance during impaired propagation conditions. There are two standard measures used to quantify the benefits provided by diversity
reception. One measure, diversity gain, specifies the reduction
in single-path impairment level (signal fading in decibels,
usually) achieved with diversity reception for a given operating time percentage (of the year or worst month). Diversity gain equals the decrease in signal-to-noise ratio, SNR
(dB), that is required to meet a given performance criterion,
relative to the SNR that would be required without diversity.
The other measure, called diversity improvement (or diversity
improvement factor or diversity advantage) is defined in the
orthogonal sense as the ratio of the nondiversity and diversity
probabilities of exceeding a specified impairment level. Both
measures are used in this article.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
DIVERSITY RECEPTION
10.0
5.0
Time percentage attenuation is exceeded
753
2.0
1.0
0.5
0.2
Path 1
0.1
0.05
Path 2
Diversity
0.02
0.01
4
6
8
10
12
11.6 GHz rain attenuation (dB)
14
one end of the link. Planning for the multiple paths is governed by the primary path impairment that the diversity configuration must overcome, such as rain attenuation, refractive
fading on satellite paths at low elevation angles, or multipath
fading on terrestrial line-of-sight links. Space diversity has
been investigated in particular for application to the mobilepropagation environment. The benefits can be very large for
the frequent and severe signal fading encountered while communicating with a terminal in motion.
In terrestrial line-of-sight telecommunications, space diversity generally refers to the use of multiple antennas displaced vertically on a tower, mainly to overcome refractivelyinduced fading caused by unwanted multipath propagation on
such links. Other examples include path diversity (access to
adjacent microwave routes in terrestrial fixed telecommunications) and site diversity (deployment of multiple earth terminals in earthspace telecommunications), both generally used
to reduce the effects of severe rainfall attenuation at frequencies above about 10 GHz.
Figure 1 shows 11.6 GHz cumulative rain attenuation distributions derived from measurements with two terminals
separated by 35 km on propagation paths at an elevation
angle of 18 (13). The curves labeled Path 1 and Path 2
are the rain attenuation distributions for the individual
paths. The Diversity curve is the joint cumulative rain attenuation distribution obtained via computer simulation by
always selecting the lesser-faded signal for each concurrent
pair of rain attenuation samples:
AJ (t) = min{A1 (t), A2 (t)}
(1)
DIVERSITY RECEPTION
12
20 GHz
16 GHz
30 GHz
10
14 dB
12 dB
10 dB
8 dB
4
6 dB
2
0
4 dB
2 dB
0
10
15
20
25
Site separation (km)
30
Single-path attenuation
15.3 GHz
35
The increase in path availability achievable by (perfect) diversity switching is observed to be considerable, although independent fading statistics (equivalent to the product of the two
single-path probability distributions) are almost never observed for dual diversity paths (14) because of mesoscale
(widespread) rainfall effects. Practical diversity-switching
systems are unlikely to achieve the degree of improvement
indicated in Figure 1 since the switching algorithm would
purposely be designed to avoid switching under conditions for
which the resulting performance gain is not required to meet
system performance objectives.
Space diversity gains are large during periods of heavy
rainfall since it is highly probable that there is considerable
spatial variability in the rain intensity. For terrestrial and
earthspace links, separating diversity antennas by several
kilometers greatly reduces the probability of simultaneous
large fades on the different paths. Antenna separation is the
dominant parameter, as illustrated in Figure 2 for several
earthspace experiments at frequencies from 15 to 30 GHz
(15). The majority of the available diversity gain is in fact
achieved with separations smaller than 10 to 15 km. At temperate latitudes, diversity gains measured during the spring
and summer (thunderstorm) seasons are generally significantly greater than those observed during the fall and winter (16).
The International Telecommunication Union (ITU) has
compiled results of many earthspace site diversity experiments, mainly in the 10 GHz to 20 GHz range, to derive an
average representation of site diversity improvement that
gives the decrease in unavailable time percentage attained by
dual-site diversity for a specified impairment level (17). The
improvement is plotted in Figure 3 for site separations of
0 km (no improvement) up to 50 km. Results in this figure
are approximate, since weaker influences such as frequency
and path elevation angle are not explicitly taken into account.
The figure indicates that performance gains achieved by increasing the site separation tapers off for separation greater
than about 1520 km. As an example, the curve for a separation of 30 km indicates that a single-path unavailability of
p1 0.01% (about one hour per year) can be reduced by two
(2)
1
0
2
5
10
20
30
50
754
10
1
5
2
km
km
km
km
km
km
km
10 2
5
2
10
3
5
2
10 4
10 1
10 2
10 1
Time percentage for a single site
DIVERSITY RECEPTION
10
Attenuation (dB)
8
Satellite 2
(18.7 GHz)
6
4
Satellite 1
(19.8 GHz)
2
0
11:10
11:30
Time (GMT)
11:50
Protection of large frequency bandwidths by transferring traffic from one satellite to another (and accommodating the corresponding network reconfiguration) appears difficult.
Angle Diversity
The general technique of configuring antennas to detect signals that have propagated along different paths, and which
appear at the receive antenna with different directions of arrival, is called angle diversity. On terrestrial paths, vertically
separated space diversity antennas may be mounted with
slightly different pointing angles with respect to the normal
line-of-sight direction to the transmit antenna to enhance reception during periods of angle-of-arrival variations caused by
refractive structures near the earth surface. Vertical beam
separation is preferable to horizontal separation in this application because variation in the refractive structure of the troposphere is more pronounced in the vertical direction. Another approach, used in troposcatter systems, is to employ a
large antenna with several feeds configured to yield somewhat different pointing directions to create some degree of
decorrelation among the corresponding propagation paths.
In troposcatter systems, the optimal vertical separation between the beams is about one antenna beamwidth, although
the range of the optimum is rather broad (3). For terrestrial
microwave paths, more complicated procedures are required
to determine the diversity gain provided by angle diversity,
which is related to the average angle of arrival as determined
by the average value of the vertical refractivity gradient for
the location of interest (4).
Polarization Diversity
For environments in which the polarization properties of a
signal are altered during propagation, orthogonally-polarized
transmission channels may become sufficiently decorrelated
for polarization diversity to be effective. Polarization diversity
is the reception of a signal on two mutually orthogonal polarizations, with or without transmission in the same two polarizations (20). If both polarizations are transmitted, a 3 dB
penalty in transmit power per channel is imposed with respect to single-polarization transmission, as the power must
755
(3)
(4a)
(4b)
756
DIVERSITY RECEPTION
(4c)
104
Channel 1
103
creases rapidly with increasing frequency. If traffic on a highfrequency channel (say, at 14/12 or 30/20 GHz), where rain
attenuation is severe, can be switched to a lower-frequency
(such as 6/4 GHz), the probability of successful transmission
is much enhanced. Frequency diversity requires that spectrum be made available at the diversity frequency to carry
the information that would otherwise be lost in the impaired
channel, and that the traffic either be satisfactorily prioritized or spare capacity held in reserve, to allow the transfer
of protected channels to the alternate satellite when required.
Diversity in this form might be used to protect smaller terminals of modest bandwidth capacity, such as Very Small Aperture Terminals (VSATs), by providing a small spare capacity
on a satellite that can be accessed by terminals when experiencing propagation impairments. However, frequency diversity is less likely to be useful for large trunking earth stations, which generally continuously utilize all of the allocated
frequency spectrum in a given band to support many communication links, and for which alternate provision will be expensive.
Time Diversity
Time diversity refers to the exploitation of the time-varying
nature of the signal impairments to retransmit information
at suitable time intervals. A commonplace analogy is resending an unsuccessfully transmitted facsimile. Information
should be retransmitted at time intervals somewhat greater
than the reciprocal of the signal fading rate, to ensure adequate decorrelation between successive transmissions (1). If
the fading conditions are quite variable, adaptive adjustment
of the time interval may be necessary to attain efficient performance over the anticipated range of environmental situations.
In digital systems, such as packet-switched networks, the
bit stream can be reconstituted from successfully received
packets, possibly by interpolation between successfully-received packets or other sophisticated methods (24), allowing
for powerful implementation possibilities. Overhead capacity
must be provided both for network control and to identify and
process the information bits, so increased spectrum and
higher data rates may be needed to maintain sufficient information throughput. In time diversity systems, storage of the
communication information is required at both the transmitter (to permit retransmission) and receiver (to support bit manipulation and message reconstruction), which constitutes an
important disadvantage for many analog, real-time, and
wideband applications.
Channel 2
RAKE Diversity
102
Frequency
diversity
101
100
106
104
102
100
Bit error rate (BER)
RAKE diversity is effective with wideband signals if the individual multipath components (echoes) can be separately identified and processed (1). RAKE diversity, also called
multipath diversity or path-delay diversity, takes advantage
of the existence of multipath components in its operation. It
can be viewed either as a variation of frequency diversity (as
in spread spectrum systems, where to reduce small-scale
multipath effects the transmission signal is spread by several
times the frequency width of the reciprocal of the delay
spread) or time diversity (where the incremental time delays
among components are used with a system time reference to
identify and process the individual multipath components).
DIVERSITY RECEPTION
757
(5)
EI2 + EQ2
(6)
758
DIVERSITY RECEPTION
r r 2 /2b
0
e
b0
r0
(7)
r 2i
2N
(8)
1 /
e i
(9)
The probability that the SNR in the ith channel does not exceed a particular threshold value of interest, S, where the
subscript S indicates selection diversity, is obtained by integrating the probability density over the domain of interest:
S
P(i S ) =
p(i )di = 1 e S /
(10)
S =
S
M
1
d
PM (S ) dS =
dS
k
k=1
(12)
(b)
(11)
8
Diversity gain (dB)
PM (t ) = (1 e S /
)M
10
(c)
6
(a)
4
3
4
5
6
7
8
Number of branches, M
10
DIVERSITY RECEPTION
759
r1(t)
r2(t)
R(t)
R0
p(R ) =
RM1 e R /
M (M 1)!
R 0
(16)
II
t1
III
t0
t2
t0
t3
series (portions prior to switching from, and portions subsequent to switching back to, either envelope) are uncorrelated,
the density function of the composite (switched) carrier envelope, R(t), can be established, as in the previous case for a
single envelope. The composite probability density is found to
be comprised of two Rayleigh densities with different
weighting factors (2).
Maximal Ratio Combining. In maximal ratio combining, the
signals are phased and coherently summed, instead of being
selected one at a time. The complete scheme is to cophase the
M channels, then apply weights, wi, to the signals that are
proportional to the SNRi of the individual channels, and finally sum these signals. The resultant envelope, r, is the sum
of the weighted envelopes ri:
r=
M
wi ri
PM (R ) = 1 e R /
(17)
R =
M
M
i =
= M
i=1
(18)
i=1
(13)
r=
i=1
M
(R /
)k1
(k 1)!
k=1
M
ri
(19)
i=1
NT = N
M
w2i
(14)
i=1
M
M (r 2 + r 2 )
r 2i
I
Q
=
R =
i =
2N
2N
i=1
i=1
i=1
(15)
For equal noise power in all diversity channels, the corresponding output SNR is E r2 /2NM, a sum of Rayleigh variables, for which there is no general solution for the probability distribution function. Solutions generated numerically
reveal that the performance of equal gain combining is only
slightly worse than maximal ratio combining (usually by less
than 1 dB).
The general expression obtained for the average value of
the output SNR is given by (2)
E =
1 + (M 1)
4
(20)
760
DIVERSITY RECEPTION
(21)
where Q(a, b) is expressed in terms of the zeroth-order modified Bessel function, I0, as
Q(a, b) =
e(a
2 +x 2 )/2
I0 (ax)x dx
(22)
2S
(1 + ||2 )
b=
2S
(1 ||2 )
(23)
Figure 8 displays the cumulative distribution functions computed with these expressions (2). The curve for 2 0 corresponds to zero correlation, as in prior analyses, while for 2
1, no diversity advantage is conferred by switching between
the two channels. However, substantial diversity gain is
achieved even when the correlation between the two signal
envelopes approaches 0.8, attesting to the efficacy of diversity
operation for this environment. Similar results are obtained
for other modes of diversity combining.
30
40
50
60
70
80
90
95
2 = 1.0
98
99
0.8
99.5
0.6
99.8
0.3
99.9
99.95
99.98
99.99
25
20
15
10
Mean SNR (dB)
M1
M1
1
2 M1 R /
pM (R ) = (1 )
e
n
n=0
n
(24)
1
R 2
(1 2 ) n!
The first term after the summation is the binomial distribution, and is the correlation coefficient between the pilot and
the adjacent signal channel. (Note that here correlation is desirable, unlike the case for envelope correlation.) The first moment (mean) of R is obtained by integrating with respect to
the probability density in Eq. (24):
R =
R pM (R )dR =
[1 + (M 1) 2 ]
(25)
0
and is equivalent to the mean SNR. The probability distribution obtained by integrating over the density function is
R
PM (R ) =
pM (x) dx
0
M1
n
M1
(R /
)k
R /
=1e
2n (1 2 )Mn1
n
k!
n=0
k=0
(26)
which represents the statistics of the combiner output signal.
DIVERSITY RECEPTION
PM (R ) = 1 e R /
M1
k=0
(R /
)k
k!
R = M
(27)
R =
(28)
10
20
30
40
50
60
70
90
95
2 = 0
5
0.7
99.5
1.0
0.5
99
99.8
99.9
99.95
99.98
99.99
30
20
Other System Considerations. Even if diversity operation offers substantial improvement in overall circuit availability,
other practical constraints may need to be taken into account.
For example, site diversity significantly improves availability
for earthsatellite paths subject to rain attenuation, as verified in Fig. 3. However, in large earth stations that utilize
wide bandwidths to serve many users, potential outages related to switching among available diversity signals is a severe problem to be avoided. Therefore, the entire receive band
for the diversity channel must be conditioned and synchronized with the main-station signal to support switching
among channels with no loss of information.
However, on the uplink to the satellite, such synchronization is extremely difficult due to variations in radio path
length (such as caused by satellite motion) between the earth
stations and the satellite. Therefore, uplink site diversity is
much less viable than downlink site diversity, except possibly
in packet-switched applications where lost packets can be recovered. A potential compromise solution for this case is to
protect the downlink path with site diversity, but implement
transmitter power control (27) to increase the availability of
the uplink path.
Figure 10 shows a 14/11 GHz earth-space site diversity
configuration (28), planned for the two sites represented in
Fig. 1. In this system, the entire downlink receive band (500
MHz) from the secondary station is transported to the main
station by the microwave Diversity Interconnect Link (DIL),
buffered and synchronized with the main station receive signal, and made available at the diversity switch. (Signal combining of the diversity signals is unlikely to be considered for
this wideband application because of the difficulty in matching phase variations across the two 500 MHz receive bands.)
Signal regeneration (demodulation and remodulation) is implemented in the DIL, not only to support frequency conversion, but also to preserve the quality of the transmissions. In
this design, the uplink signal is also made available at both
transmit sites, but this capability mainly increases the reliability of the overall system by enabling a redundant uplink
signal transmission capability.
RECENT DEVELOPMENTS
98
0.2
80
761
10
Mean SNR (dB)
10
Figure 9. Rayleigh-fading statistics for 4-channel maximal ratio diversity combining with combiner errors (specified by 2) between reference pilot and channel signal. Fade distributions are referenced to
the SNR for a single (nondiversity) channel. ( 1994 IEEE.)
762
DIVERSITY RECEPTION
14 GHz
11 GHz
14 GHz
11 GHz
Diversity switch control
TX
UC
RX
11 GHz
4 GHz
RX
LNA
HPA
14 GHz
IPA
HYB
HPA
DC
IPA
6 GHz
TX
DC
LNA
UC
RX
TX
TX
UC
4
GHz
6
GHz
UC
Rx
Customer interface
Main site
Repeater
Diversity site
Figure 10. Configuration for 14/11 GHz earthspace site diversity, including microwave link
that interconnects the main and diversity stations to support selection diversity (TX/RX
transmit/receive; IPA/HPA intermediate/high-power amplifier; LNA low-noise amplifier;
UC/DC up/down converter; HYB hybrid). (Copyright 1979 COMSAT Corp. All rights reserved by COMSAT Corp. Used by permission.)
99
98
Time percentage fade is exceeded
Highest satellite
Best satellite
2 best satellites
3 best satellites
4 best satellites
95
90
80
70
60
50
40
30
20
10
5
2
1
5
10
15
20
Signal fade (dB)
25
30
35
DIVERSITY RECEPTION
763
BIBLIOGRAPHY
5
4
1. M. Schwartz, W. R. Bennett, and S. Stein, Communication Systems and Techniques (in particular chaps. 10 and 11), Piscataway,
NJ: IEEE Press, 1996.
2. W. C. Jakes (ed.), Microwave Mobile Communications, Piscataway, NJ: IEEE Press, 1994.
1
0
10
15
20
25
Cross coupling (dB)
30
5. D. L. Bryant and J. E. Allnutt, Use of closely-spaced height diversity antennas to alleviate the effects of low angle non-absorptive
fading on satellite slant paths, Electron. Lett., 26: 480481, 1990.
6. S. H. Lin, T. C. Lee, and M. F. Gardina, Diversity protections
for digital radiosummary of ten-year experiments and studies,
IEEE Commun. Magazine, 26 (2): 5164, 1988.
7. R. G. Vaughan and J. Bach Andersen, Antenna diversity in mobile communications, IEEE Trans. Veh. Technol., VT-36: 149
172, 1988.
There will be many instances when at least one of the satellite paths is free of obstruction. If the most favorable path
can be selected as required, or the multiple signals can be
combined, path performance is enhanced. Figure 11 shows
the cumulative fading distributions estimated by simulating
the path availability achieved by equal-gain combining of
available signals from multiple satellites for an urban
shadowing/blockage environment (33). The substantial benefits of path diversity for this environment are apparent. In
such systems using path diversity, the RAKE receiver technique may be implemented, not only to overcome shadowing
and blockage impairments, but also to enable smooth handoff
among the available satellite beams (34), which might be required every few minutes in LEO systems, since satellite motion is rapid with respect to earth-based terminals.
Polarization diversity has found renewed interest in mobile telephony as a means to improve performance while
avoiding the need for additional spectrum or a second spacediversity antenna, especially for handheld terminals (35) in
which the antenna orientation is quite variable. In a study of
polarization-diversity reception based on measurements in an
urban area (36), the achievable diversity gain was related to
the cross-polarization discrimination, XPD, of the received
signals. Figure 12 shows diversity-gain results at the 90%
probability level for selection combining with worst-case reception (assuming the two signals arrive 180 out of phase),
referenced to the 45 branch of two linearly-polarized elements oriented at 45. The results indicate that the diversity gain achievable with polarization diversity in this application is only a dB or so less than typical space diversity gains
achieved with antenna separations of the order of 20 wavelengths. In this environment, better performance was obtained with linear polarizations oriented at 45 than for vertical and horizontal polarizations, as the former maintained
more nearly equal mean signal levels in the two diversity
channels.
Additional diversity reception applications will inevitably
arise in response to future developments in telecommunication systems and technology.
8. V. Mimis and A. Smalley, Low elevation angle site diversity satellite communications for the Canadian arctic, Conf. Record, IEEE
Int. Conf. Commun., Philadelphia, 2: 4A.4.14A.4.5, 1982.
9. C. Kittiver and E. Sigler, Propagation considerations in the
American Mobile Satellite System design, Proc. 17th NASA Propag. Experimenters Meeting, Pasadena, CA: JPL Publ., 9321:
171181, 1993.
10. J. C. Cardosa, A. Safaai-Jazi, and W. L. Stutzman, Microscale
diversity in satellite communications, IEEE Trans. Antennas Propag., 41: 801804, 1993.
11. A. Bosisio et al., Analysis and application of short-distance site
diversity techniques for 20/30 GHz communication links, Conf.
Record, IEEE Global Telecommun. Conf. : Singapore, 1: 749
753, 1995.
12. D. C. Cox, Universal digital portable radio communications, Proc.
IEEE, 75: 436477, 1987.
13. D. V. Rogers and G. Hyde, Diversity measurements of 11.6-GHz
rain attenuation at Etam and Lenox, West Virginia, COMSAT
Tech. Rev., 9: 243254, 1979.
14. D. V. Rogers, Diversity and single-site radiometric measurements
of 12-GHz rain attenuation in different climates, IEE Conf. Publ.,
No. 195, Part 2: 118123, 1981.
15. D. B. Hodge, An empirical relationship for path diversity gain,
IEEE Trans. Antennas Propag., AP-24: 250251, 1976.
16. J. Goldhirsh et al., Three-site space-diversity experiment at 20
GHz using ACTS in the eastern United States, Proc. IEEE, 85:
970980, 1997.
17. Recommendation ITU-R P.618-5, Propagation data and prediction methods required for the design of earth-space telecommunication systems, International Telecommunication Union, Geneva:
ITU-R Recommendations 1997 P Series, Part 2: 305323, 1998.
18. D. B. Hodge, An improved model for diversity gain on earth
space propagation paths, Radio Sci., 17: 13931399, 1982.
19. E. Matricciani and M. Mauri, Italsat-Olympus 20-GHz orbital diversity experiment at Spino dAdda, IEEE Trans. Antennas Propag., 43: 105108, 1995.
20. R. G. Vaughan, Polarization diversity in mobile communications,
IEEE Trans. Veh. Technol., 39: 177186, 1990.
21. P. C. F. Eggers, J. Toftgard, and A. M. Opera, Antenna systems
for base station diversity in urban small and micro cells, IEEE J.
Sel. Areas Commun., 11: 10461057, 1993.
764
DIVERSITY RECEPTION
DAVID V. ROGERS
Industry Canada
487
488
Er
Hr
Ht
z
Diffuse Scattering
The laws for specular scattering are only valid for electrically
large scattering bodies. If the object causing the electromagnetic scattering is small compared to the wavelength of the
incident electromagnetic field, the induced currents would
tend to bend around the contour, creating diffuse scattering.
Unlike specular scattering, diffuse scattering results when
the electromagnetic energy spreads out as it radiates from the
scattering object. The smaller the object, the more the energy
will spread as it reradiates.
A simple example of diffuse scattering is an electromagnetic field incident upon a cylindrical conductor of small radius, where the electric field is parallel to the axis of the cylinder. At frequencies below the microwave region of the
electromagnetic spectrum, a thin copper wire will have an
electrically small radius. A first approximation is to assume
that the current that is induced by the incident field is uniformly distributed across the entire cross section of the wire.
The scattered or reradiated field from this wire is similar to
the field radiated from a wire antenna having the same linear
current distribution. However, the total field around the wire
is the superposition, or vector sum, of the scattered field and
the original incident field that would have existed without the
wire present. The scattered field, added to the incident field,
creates a pattern with constructive reinforcement in some directions and destructive cancellation in other directions. This
is the function of the passive elements found on the YagiUda
antenna, common in television and other VHF and UHF communications.
In the YagiUda antenna, only one set of elements is active. The active (or driven) element usually makes up a halfwavelength dipole antenna. The other elements are simply
conductive cylinders or wires that reradiate some of the energy incident on them from the active element. Depending on
the relative lengths of these passive elements, each of their
radiated fields will add to the incident field of the active element, to create an overall pattern of power flow (1) This focusing of energy is called antenna gain. Other forms of diffuse
scattering by electrically small bodies are not so intentional.
Et
2, 2, 2
Ei
1, 1, 1
Hi
Figure 1. Illustration of Snells law. A plane-wave electromagnetic
field obliquely incident onto a plane boundary, separating medium 1
(1, 1, 1) from medium 2 (2, 2, 2). A reflected field and a transmitted field scatter from this discontinuity in electrical constants.
(1)
Diffuse Diffraction
Another form of scattering, which cannot be accounted for by
reflection or refraction, is diffraction. For electrically large
scattering bodies, diffraction appears to occur at geometrical
discontinuities such as edges and corners. A first approximation is that currents induced only at these discontinuities reradiate electromagnetic energy. Diffraction is the scattering
mechanism that accounts for radiation filling in the region
that would have been completely blocked (shadowed) by an
opaque obstruction.
Diffraction and reflection are important scattering mechanisms in communications. Whether in an urban or rural envi-
xi + yyE
yi )e j 1
(xxE
sin i x
xr + yyE
yr )e j 1
+ (xxE
sin r x
= (xxE
xt + yyE
yt )e j 2
sin t x
(2)
This equality can only be true for all x when the exponents,
or phases, are equal:
1 sin i x = 1 sin r x = 2 sin t x
(3)
Equation (3), proves the well-known specular law for reflection (5),
i = r
(4)
r
2 2
(5)
1 1
489
2. Region 2 represents a scattering body of material consisting of a different permittivity , permeability , and conductivity , than those of the surrounding region 1. Region 2
is bounded by a surface S. Assume that an electric field, E1,
and its associated magnetic field H1, originate from a source
current density J0. This source could simply be the current
oscillating in a transmitting antenna. These fields propagate
undisturbed, through region 1, until they become incident
upon the scattering body of region 2. As the fields cross the
boundary between region 1 and region 2, they will be perturbed, that is, E2 and H2 in region 2 will generally not be
equal to the fields E1 and H1 propagating in region 1. This
abrupt change or discontinuity in electric and magnetic field
strength results in currents that are induced at the discontinuity. In general these currents will be distributed through
the volume of regions 1 and 2, depending on their electrical
constants.
According to the induction theorem, whenever there is a
discontinuity of the E and H fields crossing a boundary S between two media with different electrical constants, one can
assume that induced currents at S cause the discontinuities
in the fields. The induced current can be an electric current
sheet (68)
ELECTROMAGNETIC THEOREMS
Many electromagnetic scattering problems do not lend themselves to the simple application of the laws of reflection and
refraction. To develop more sophisticated analysis tools, a discussion of some basic electromagnetic theorems will be useful.
J s = n (H
H s H t ) = n H i
Uniqueness Theorem
Knowledge of the sources induced on the surface of a scattering body S enables unique solutions of the fields reradiated
by those induced sources. Conversely, the known fields allow
a unique calculation of the induced sources. The electric field
E and magnetic field H are uniquely determined if (6,7)
1. n
E, the tangential component of E, is specified on
S,
2. n
H, the tangential component of H, is specified on
S, and
3. n
E is specified on part of S, and n
H is specified
on the remaining part of S.
Induction Theorem
In general, sources, such as conduction, displacement, and polarization currents, are induced at electrical discontinuities in
the medium through which the incident field is propagating.
Figure 2 shows a typical discontinuity represented by region
J0
E 1, H 1
JS
MS
2, 2,
^
n
1, 1,
(6b)
(7a)
M s = n E i = n E s
(7b)
and
E 2, H 2
M0
(6a)
(8a)
490
and
Wavefront
M s = n E s = n E i
(8b)
z
r
GO shadow
DIFFRACTION
Diffraction is the scattering mechanism that neither reflects
off nor transmits through a structure. Even with opaque
structures, allowing no transmission, diffraction accounts for
radiation into the geometrical shadow region. This scattering
mechanism cannot be modeled with Snells law. To analyze
diffraction exactly would require more knowledge about the
induced current distribution around the scattering structure
than would typically be available. Therefore approximations
must be made to simplify the analysis. Two common approaches to analyzing diffraction are the use of geometrical
optics and physical optics.
Conducting
half plane
Figure 3. Half-screen diffraction using the GTD. The currents induced at the edge of the conducting half screen radiate into the GO
shadow region.
Geometrical Optics
Geometrical optics (GO) is a ray-tracing technique that assumes that the electromagnetic energy travels in straight
parallel lines, or rays, that are perpendicular to the wavefront. These rays travel from the point of reradiation to the
observation point. While relatively easy to implement (9,10),
GO is an approximation that relies on some important assumptions, primarily that the wavelength of the electromagnetic field must approach zero. Clearly, GO is an asymptotic
technique only valid for sufficiently high frequencies, such
that the wavelength is small compared to the dimension of
the obstruction. Since GO assumes infinite frequency, it ignores the wave nature of the electromagnetic scattering field,
thus ignoring diffraction. The GO model creates an abrupt
change in energy at the transition from the illuminated region to the shadow region. The abrupt change in field
strength, without currents or charges to account for this discontinuity, violates boundary conditions. Therefore, GO provides only the crudest model, accounting only for reflection
and refraction, but not for diffraction.
The geometrical theory of diffraction (GTD) extends GO to
account for diffraction, by introducing a diffraction coefficient, D, analogous to for reflection and to T for transmission (1113). The total electric field ET around the obstruction
is
ET = Eg + Ed
(9)
(10)
12
r j r
e
E0
x
(11)
Wavefront
Secondary
sources
E 0 j (r+ )
E=
e
dE
r+
(12)
r
x0
where r is the distance from the observation point to the conducting half plane, and is the additional distance to the secondary sources. From Fig. 5,
(r + )2 = r2 + 2r + 2 = x2 + r2
GO shadow
Conducting
half plane
(13)
Eg + Ed with E0 = 1 : z = 2 wavelengths
491
x2
2r
(14)
1.5
E = E 0 /r e j r
0.5
0
e j x
2 /2r
dx
(15)
x0
0.5
1
20
E=
15
10
5
0
5
x (wavelengths)
(a)
10
15
20
2r
E 0 e j r
e j u
2 /2
du
(16)
u0
Eg + Ed with E0 = 1 : z = 20 wavelengths
E=
2r
E 0 e j r
e j u
2 /2
uo
du
e j u
2 /2
du
(17)
1.5
1
0.5
E=
2r
E 0 e j r
1
2
+ j 12 [C(u0 ) + jS(u0 )]
(18a)
0
0.5
1
20
15
10
10
15
20
x (wavelengths)
(b)
Figure 4. Simulation of half-screen diffraction by the GTD. The solid
straight line represents the GO incident field. The oscillating curve
is the diffracted field, calculated by the GTD, added to the GO field,
at a distance behind the screen of (a) z 2 wavelengths and (b) z
20 wavelengths.
where C(u0) and S(u0) are the Fresnel sine and cosine integrals respectively (1921). The solution to Eq. (18a) is similar
to the GTD solution for Fig. 3, with the exception that Eq.
(18a) for PO does not suffer the discontinuity of Eq. (11) for
GTD. In fact, Eq. (18a) has an analytic solution in the GO
illuminationshadow transition region. The total electric field
at x0 0 is
E (x0 = 0) =
2r
E 0 e j r
1
2
+ j 12
(18b)
492
dx dz
E = j
J
4
R
S
1
4
S
e j R
dx dz
M
R
4
M
S
1
4
e j R
dx dz
R
S
e j R
dx dz
J
R
E
H = jE
(19b)
where the primed symbols refer to the source rather than the
field. The equivalence principle allows the electric field in the
aperture to be replaced by the magnetic current sheet Ms over
a continuous conducting screen (7). The aperture is essentially shorted, which cancels Js. From image theory, it appears as though an identical image of Ms lay on the opposite
side of the screen. Since these two current sheets nearly coincide, the entire problem can be replaced with 2Ms at the aperture location, and no screen at all. Then Eq. (19b) becomes
e j R
dx dz
H = j
M
(20)
2
R
(21)
(22)
(19a)
and
H=j
Rather than solving the integrodifferential Eq. (19a), Amperes law can be used to obtain directly the scattered electric
field in the source-free region,
2D2
(23)
H = j
E
2r 0
a/2
a/2
b/2
e j (x
dx dz
b/2
(24)
While appearing messy, Eq. (24) is a straightforward integral. After integrating the two exponential terms, substituting the limits, and applying the identity
sin =
S
e j e j
j2
H= j
abe j r
r
sin X sin Y
X
(25a)
where
R
X =
a sin cos
2
(25b)
Y=
b sin sin
2
(25c)
and
y
Figure 7 is a plot of Eq. (25a), with the amplitude normalized. The x dimension is a 6, the y dimension is b 3,
and the observation screen is z 100 from the aperture.
The normal incidence was chosen for this problem to illustrate the concept while keeping the mathematics simple.
493
v
E( m )
1
0.8
0.6
0.4
0.2
0
10
5
y( )
10
0
5
x( )
5
10
5
10
(26)
(27)
The first term in Eq. (27) is the ratio of the field diffracted by
the aperture to the field with no screen present at all, and the
second term is the ratio of the field produced by the complementary screen to the conjugate source. The conjugate source
refers to the opposite field rotated by 90. In vector form, Eq.
(27) can be rewritten as
E c = E i H a
(28)
Figure 7. Normalized scattering pattern through the aperture. The dimensions of the aperture are 6 by 3 wavelengths;
the observation screen is 100 wavelengths from the plane of
the aperture. The scattering pattern is wider in the x direction, since the x dimension of the aperture is twice the y dimension.
which indicates that the electric field scattered from a conductive plate can be calculated from the field scattered from the
aperture, by subtracting the latter from the incident field (22).
SPECIAL CASES OF ELECTROMAGNETIC WAVE SCATTERING
Rayleigh Scattering
If the scattering object is much smaller than a wavelength,
its scattered energy varies inversely as the fourth power of
the wavelength (1,24,25). Therefore, for a given subwavelength object, higher frequencies will scatter more than lower
frequencies. This is the basis behind the concept of Rayleigh
scattering for small scatterers. In fact, Rayleigh scattering
answers the commonly asked question: Why is the sky blue?
Since the blue end of the visible spectrum has the shortest
wavelength, blue light scatters more than the rest of the visible spectrum from dust, water, and even air molecules. As the
scattering objects become larger, they fall into the category
called Mie scattering.
Radar Cross Section
Electromagnetic wave scattering is the basis by which radar
signals are returned to the radar receiver from a target. Since
the typical radar system employs a colocated transmit and
receive antenna, the source and observation points are the
same. This scenario is a specific case of electromagnetic wave
scattering, as previously discussed, and is known as monostatic scattering.
As the transmitted power Pt propagates through space, it
spreads over an increasing surface area, A, resulting in decreased power density St Pt /A. If Pt spreads spherically, as
with a point source or isotropic radiator, A 4d2, where d
is the distance from the transmitter. A target can intercept
part of the transmitted power and scatter it in various directions. The radar cross section (RCS) is the effective area of
the target that would return the monostatic power density
back to the source, if this target scattered the power isotropically (1,26,27). The RCS is related to the physical cross-sectional area of the target but also depends on factors such as
the frequency and polarization of the radar signal as well as
the targets shape, material, and orientation to the transmitter.
494
SUMMARY
Since electromagnetic scattering perturbs the incident field,
it can create interference, both constructive and destructive.
The three main mechanisms of scattering are reflection, refraction, and diffraction. Reflection and refraction are the
most common, and since these scattering mechanisms tend to
be specular, they are easiest to analyze. Diffraction is much
more difficult to analyze, and is typically not as dominant as
the other two.
Various approximations can lead to solutions of diffraction
problems. The two main approximations covered were the
GTD, which takes advantage of ray tracing, and PO, which
relies on the Huygens secondary sources. Slight modifications
to the PO solution for the aperture problem can lead to solutions for backscattering and forward scattering from a conductive plate. This simple structure can serve as a building
block for more complicated structures, which can be modeled
as composites of conductive plates.
BIBLIOGRAPHY
1. J. D. Kraus, Antennas, 2nd ed., New York: McGraw-Hill, 1988.
2. C. R. Paul and S. A. Nasar, Introduction to Electromagnetic
Fields, New York: McGraw-Hill, 1987.
3. C. T. A. Johnk, Engineering Electromagnetic Fields and Waves,
New York: Wiley, 1988.
4. G. G. Skitek and S. V. Marshall, Electromagnetic Concepts and
Applications, Englewood Cliffs, NJ: Prentice-Hall, 1982.
5. P. A. Tipler, Physics, New York: Worth, 1976.
6. C. A. Balanis, Advanced Engineering Electromagnetics, New
York: Wiley, 1989.
7. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, Englewood Cliffs, NJ: Prentice-Hall, 1968.
8. R. F. Harrington, Time-Harmonic Electromagnetic Fields, New
York: McGraw-Hill, 1961.
9. G. E. Corazza et al., A characterization of indoor space and frequency diversity by ray-tracing modeling, IEEE J. Selected Areas
Commun., 14 (3): 411419, 1996.
10. M. Kimpe, V. Bohossian, and H. Leib, Ray tracing for indoor radio channel estimation, IEEE Proc. 2nd Int. Conf. on Universal
Personal Commun. (ICUPC), October 1993, pp. 6468.
11. J. B. Keller, A geometric theory of diffraction, in L. M. Graves
(ed.), Calculus of Variations and its Applications, Proc. Symp.
Appl. Math., Vol. III, New York: McGraw-Hill, 1958, pp. 2752.
12. G. E. Athanasiadou, A. R. Nix, and J P. McGeehan, Indoor 3D
ray tracing predictions and their comparison with high resolution
wideband measurements, Proc. IEEE 46th Veh. Tech. Conf., April
1996, Vol. 1, pp. 3640.
13. O. Landron, M. J. Feuerstein, and T. S. Rappaport, A comparison
of theoretical and empirical reflection coefficients for typical exterior wall surfaces in a mobile radio environment, IEEE Trans.
Antennas Propag., 44: 341351, 1996.
14. R. H. Clarke and J. Brown, Diffraction Theory and Antennas, New
York: Wiley, 1980.
15. J. D. Kraus, Electromagnetics, 4th ed., New York: McGraw-Hill,
1992, pp. 622627.
16. R. G. Kouyoumjian and P. H. Pathak, A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface,
Proc. IEEE, 62: 14481461, 1974.
17. J. B. Keller, Geometrical theory of diffraction, J. Opt. Soc. Amer.,
52 (2): 116130, 1962.
RANDALL L. MUSSELMAN
United States Air Force Academy
HELICAL ANTENNAS
HELICAL ANTENNAS
The helix antenna has a long and fascinating history. It was
discovered in 1946 by John Kraus. Since then, new variations
have continued to arise, even up to the present day. The discovery itself is a very interesting story, which is told in John
Krauss book Antennas (1). It all started with an afternoon
lecture at Ohio State University. John Kraus listened as the
speaker described the wave-guiding helix used in travelingwave tubes; he wondered if this interesting helix could be
used as an antenna and, after the talk, he asked the wellknown lecturer. The speaker replied emphatically, No, Ive
tried it and it doesnt work. That very evening, John Kraus
went down to his basement, wound a seven-turn helical coil
one wavelength in circumference and fed it by means of a
coaxial line and ground plane [Fig. 1(a)]. He found that it
produced a sharp beam of circularly polarized radiation off
the open end of the helix. So the helix antenna was born,
despite the advice of experts.
plane parallel to the axis, for example, any plane that includes the axis.
Figure 1(a) shows the wire helix antenna. Parameters are
defined as follows:
D is the diameter of helix, which is equal to 2a.
C is the circumference.
S is the spacing between turns.
is the pitch angle.
L is the length of one turn.
N is the number of turns.
C, S, and L represent the respective distances in wavelengths such that C C/ , and so on. When one turn of the
helix is unrolled on a flat plane, the relationships between the
spacing S, circumference C, and turn length L and pitch angle
can be obtained from the triangle shown in Fig. 1(b) as follows:
S = C tan
Helical Curves
The helix was well-known in ancient Greece. Geminus described it in the first century B.C. and there are references to
earlier work on the helix. The cylindrical helix may be defined
by considering a right circular cylinder of radius a, whose axis
is the z axis. Using a right-handed cylindrical coordinate system (r, , z) the equations of the helix are
x = a cos
y = a sin
z = a tan
(1)
z
D
S
S
C = D
(b)
Ground
plane
Coax
(a)
Figure 1. The helical antenna. (a) A helix fed by a coaxial line and
a ground plane. (b) One turn of the helix unrolled on a flat plane.
661
= tan1
S
C
L=
C2 + S2
(2)
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
662
HELICAL ANTENNAS
antenna, implying that it is a monofilar helix. The helix commonly operates in two different modes, the normal mode and
the axial mode depending on the electrical size of the helix.
When the dimensions of the helix are small compared with a
wavelength (D , NS ), the maximum radiation is normal (or perpendicular) to the helix axis. This condition is
called the normal mode. When the helix circumference is on
the order of one wavelength, the maximum radiation is along
the helix axis. Thus, this type of operation is called the axial
mode. The axial mode helix is a broadband antenna. The radiation from this axial mode helix is close to circular polarization along the axis. There is also a backfire mode which is
discussed in a later section.
E = jIS
(a)
(b)
(c)
Figure 2. The normal mode helix. (a) Coordinate system. (b) Loop
and dipole model. (c) Beam pattern.
ejkr
sin
4r
(4)
ejkr
sin
4r
(5)
(6)
which is the same as that of the Hertzian dipole and the small
loop and is shown in Fig. 2(c). The field is zero along the axis
(in the end-fire direction) and is maximum in the xy plane
( 90), which is normal to the helix axis.
Because E and E are 90 out-of-phase, as shown in Eqs.
(3) and (4), the radiated wave is elliptically polarized. The
axial ratio (AR) of the polarization ellipse of the far field is
obtained by dividing the magnitude of Eq. (4) by that of Eq.
(3):
AR =
(3)
where A D2 /4 is the area of the loop, k the propagation constant, and / the intrinsic impedance. The
far field of the short dipole, or the Hertzian dipole of length
S, has only an E component, given by
ejkr
sin
4r
|E |
=
|E |
S
S
2S
=
=
2
2A
(D)2
A
k
(7)
C = D = 2S or C = 2S
(8)
Under this condition the radiation field is circularly polarized
in all directions except of course along the axis where the radiation is zero. The polarization ellipse of the radiation from
a helix of constant turn-length (L) changes progressively as
the pitch angle is varied. When tan1(S/C) 0 (the
helix reduces to a loop), AR 0, E 0, E a E ; thus the
wave is linearly polarized with horizontal (or perpendicular)
polarization. As increases, the polarization becomes elliptical with the major axis of the ellipse being horizontal. When
reaches a value such that condition (8) is satisfied, AR 1,
and the polarization is circular. With the help of Eq. (2), the
condition (8) leads to the following value of :
!
"
1 + L2
1 1 +
(9)
CP = sin
L
HELICAL ANTENNAS
As increases further, the polarization again becomes elliptical with the major axis being vertical. Finally, when 90
(the helix reduces to a dipole), AR , E 0, E aE;
thus the polarization is linear with vertical (or parallel) polarization. For small pitch angles ( 1), Eq. (9) is simplified to
CP = C /2
(10)
where CP is in radians. For small pitch angles, circular polarization can occur at frequencies such that the circumference
is very small compared to a wavelength (C 1).
From Eqs. (3) and (4), we note that the loop field E and
the dipole field E, respectively, are proportional to the second
and first powers of frequency. Correspondingly, radiation resistance of loop and dipole are proportional to the fourth and
second powers, respectively. Thus, as frequency decreases,
the dipole radiation predominates and the beam pattern is
linearly polarized. In this linearly polarized frequency range,
the normal mode helix has some interesting properties. Its
beam pattern is essentially that associated with the dipoles,
that is, a monopole of length NS above a ground plane. Its
impedance, however, is significantly affected by the loops.
The normal mode helix is limited by its size. It has the
same restrictions and limitations that apply to any electrically small antenna. But within those restrictions, it has certain advantages over a dipole antenna of the same height.
These include a lower frequency for resonance and a larger
radiation resistance, both because of the longer path of the
helical structure. While the dipole may require additional impedance-matching circuits to achieve resonance, the helix is
resonant without supplementary matching elements. Another
advantage over the dipole is that the helix is flexible and
more resilient. The higher radiation resistance, resonant
characteristic, and flexibility make the normal mode helix
suitable for small antennas used in mobile communications.
AF =
3
4
< C <
4
3
(11)
The bandwidth ratio, the ratio of the upper and lower frequencies is 4/3 3/4 or 1.78, which is close to a 2 : 1 bandwidth. The helix is usually fed axially or peripherally with
the inner conductor of the coaxial line connected to the helix
and the outer conductor attached to the ground plane. The
(12)
663
sin[(N/2)]
N sin(/2)
(13)
and
= kS cos +
(14)
m is an integer
(15)
(16)
664
HELICAL ANTENNAS
1
1
2 S 2N + 1
=
kS + 2 1 +
=
+
(17)
L
L
2N
L
2N
L
k
p= =
S + (2N + 1)/2N
(18)
Circular Loop Radiation. In this section we consider the radiation from a circular loop carrying a current I(). The result
is useful in understanding the operation of the helix in both
the mono- and multifilar forms. In addition, it will yield an
approximate element factor for a single turn of the helix. The
loop of radius a is centered at the origin and lies in the xy
plane. The current distribution I() may be represented in
terms of a complex Fourier series representation:
I() =
In ejn
(20)
n=
An =
A n =
e
4r
(19)
(21)
(22)
ejn ejkr In a
4r
2
0
ejn
e j + e j
2
(24)
A n =
(25)
(26)
The total fields may of course be obtained by adding contributions of all Fourier modes.
Now lets evaluate the far fields along the z axis ( 0,
180) for each of the separate Fourier modes. We note that
Jn(0) 0 (n 0) and J0(0) 1. Evaluating the cases n 1,
HELICAL ANTENNAS
(27)
I( ) = I0 e j = I0 e j a
(28)
(29)
665
(30)
(31)
K
B
[degrees]
C NS
(32)
666
HELICAL ANTENNAS
C = 0.66
275 MHz
C = 0.73
300 MHz
C = 0.85
350 MHz
C = 0.97 C = 1.09
400 MHz 450 MHz
C = 1.22
500 MHz
C = 1.35
550 MHz
= 14
n=6
Figure 3. Measured beam patterns of the monofilar axial mode helix. From Kraus (1). 1988
by McGraw-Hill, Inc. Reprinted with permission of the McGraw-Hill Companies.
G = KGC2 NS
(33)
180
150
AR =
Axial ratio
2N + 1
2N
( = 0 )
(34)
100
If N is large, the axial ratio approaches unity and the polarization is nearly circular. For example, for a six-turn helix,
AR 13/12 1.08 according to Eq. (34). This axial ratio is
independent of frequency or circumference. In Fig. 4(b), the
measured values of the axial ratio for the six-turn, 14 axialmode helix are plotted as a function of the circumference
(C). We observe that AR is nearly 1 over the range of about
0.73 C 1.4. The sense of circular polarization is determined by the sense of the helix windings.
Input Impedance. The input impedance of the axial mode
helical antenna is nearly purely resistive. The empirical formulas for the input resistance are given (1) by
50
0
6
5
4
3
2
1
7
6
VSWR
where KG is the gain factor which depends on the design parameters. King and Wong (7) report that KG varies from 4.2
to 7.7. Experiments show that the gain is peak when C is
slightly larger than 1.
Axial Ratio. We have shown from the approximate analysis
described in a previous section that the radiation field is circularly polarized in the mainbeam direction ( 0), implying
AR 1. With a more accurate analysis including the effect
of relative phase velocity for increased directivity, Kraus (1)
obtains the axial ratio along the helix axis as follows:
+
X
5
4
3
2
1
.5 .6
.7
.8
Rin = 140C
(35)
Rin = 150/ C
(36)
within 10% for the case of peripheral feed. Both relations are
valid when 0.8 C 1.2, 12 14 and N 4. With a
suitable matching section, Rin can be made any desired value
from 50 to 150. In the inset of Fig. 4(c), trends of input
resistance R and reactance X are shown as a function of the
relative frequency or circumference. Note that R is relatively
constant and X is very small for 0.7 C 1.5. Figure 4(c)
also shows the voltage standing wave ratio (VSWR) measured
on a 53 coaxial line. We observe that the VSWR nearly re-
;;
;;
;;
;;
;;
;
;;;;;
;;;;
Ends
open
HELICAL ANTENNAS
667
Ends
shorted
Broadband Characteristics. Considering all the characteristics of beam pattern, input impedance, and polarization as a
function of circumference, we find that the performance of the
axial mode helix is satisfactory over the range of about
0.75 C 1.25 within the restrictions given on and N.
Thus the bandwidth, defined by the ratio of upper and lower
frequencies, is almost an octave. The broadband characteristics of the helix can be explained by the natural adjustment
of the phase velocity. As the helix size C, or equivalently the
frequency, varies over rather wide range, the phase velocity
adjusts itself automatically such that the fields from each
turn add nearly in phase in the axial direction.
Tape 4
Tape 3
A slight taper on the end of the helix (9,10) reduces the axial
ratio at the expense of a slight reduction in gain. Axial ratio
is improved both on and off axis. A taper is also used at the
input to improve impedance characteristics. A circular-cavity
backing is sometimes used to reduce the back radiation and
increase the forward gain. Dielectric-tube support has been
used with the helix antenna. This lowers the frequency for
the onset of axial-mode operation and has an effect on the
terminal impedance. A solid dielectric core has also been used
with the helix (the polyrod helix). A helix with an inner concentric metal core has been used as a TV transmitter (11).
The antenna utilizes higher order Fourier modes such as
ej2 , ej5 which radiate sidefire rather than endfire. This is
particularly useful with towers and masts whose circumference is much larger than a wavelength. An array of helices is
stacked along the mast to produce the required beam pattern.
The helical antenna has often been used as an element in
various types of arrays. Large planar arrays of helices have
been used in radio astronomy (12). An array of axial-mode
helices has been used for global positioning system (GPS) satellite transmitters (1). Helices are also used as feeds for parabolic dishes. Applications of the helix are legion.
MULTIFILAR HELIX ANTENNAS
Ground
plane
Tape 2
Tape 1
DG
(a)
Feed region
(phases 0, 90, 180, 270)
(b)
668
HELICAL ANTENNAS
270
270
180
180
90
90
(a)
(b)
270
270
180
180
90
90
(c)
(d)
270
270
180
180
90
90
(e)
(f)
270
270
180
90
cos
=
1 + sin
90
(g)
180
(h)
cos
1 sin
(37c)
M
cot
2
(37d)
HELICAL ANTENNAS
proximately 0.75 C 1.25 as noted previously. In comparison, Eq. (37) predicts that Cmin 0.78 and Cmax 1.28 for an
approximate bandwidth of 0.78 C 1.28. The two bandwidth ratios are very close. Equation (37) has been applied to
bifilar, quadrifilar, and octofilar axial beam helices, and the
results agree well with experiments as shown in Refs. 1517.
Figure 6 shows the beam patterns of a quadrifilar helix antenna with a ground plane [see Fig. 5(a)]. The pitch angle
is 35, diameter D is 3, antenna length is 24, ground plane
diameter DG is 10, and the tape width is 1/2. The feed system (16) provides four outputs that are phased 0, 90, 180,
and 270, each output being connected to one of the four wires
of the quadrifilar helix antenna. Equations (37b,c) yield
Cmin 0.52 and Cmax 1.92 for 35 and M 4. Thus the
bandwidth of the antenna is given by 0.52 C 1.92 for a
bandwidth ratio of 3.7 : 1. The progression of the beam patterns may be described as follows. At a frequency somewhat
below the lower limit, backfire operation begins, as evidenced
by the strong backlobe at C 0.44. The backlobe decreases
rapidly as we approach the lower limit. The axial ratio also
decreases rapidly and is less than 2 : 1 at the lower limit.
Other antenna characteristics such as VSWR are also acceptable (17). The axial mode then predominates over the 3.5 : 1
bandwidth. The beam pattern narrows steadily and the directivity increases with frequency. The upper limit occurs at
C 1.92, at a frequency just above that of Figure 6(f). Above
the upper limit, beam-splitting occurs at C 2.10 and complete pattern breakup at C 2.70. Thus, with the multifilar
helix, the bandwidth of the axial mode is extended to both
lower and higher frequencies as predicted in Eq. (37).
The beam pattern at C 1.80 exhibits the undesirable
characteristic of high side lobes. This is caused by the backfire
operation, which changes from backfire through sidefire towards endfire as frequency increases. It is sometimes called a
scanning mode. The quadrifilar helix may also be used in a
counterwound version with both right- and left-hand windings. The on-axis polarization is linear rather than circular.
The backfire mode is much more effectively suppressed in this
version, and the sidelobe levels are much lower. Bandwidths
are between 4 : 1 and 5 : 1 as shown in (16). Gerst and Worden
(15) describe a 53 pitch angle counterwound octofilar helix
with 9 : 1 bandwidth. The multifilar helix antenna does not
radiate in a normal mode at low frequencies because of the
phase excitation 0, 90, 180, and 270 of the windings. The
excitation ej is a supergain excitation at low frequencies
(C 1). Details on the axial multifilar antenna are given in
(1517) and related references.
= 0
669
90
180
Figure 7. Beam pattern of the backfire quadrifilar helix.
670
SUMMARY
The helical antenna, first discovered in 1946 by John Kraus,
has evolved into many different forms with many different
applications. The normal mode helix has some advantages for
low-frequency applications. The broadband, circularly polarized, axial mode helix radiates forward endfire along its axis.
It has been the most widely used of all forms of the helix. The
quadrifilar axial mode helix extends the bandwidth further.
The backfire quadrifilar helix radiates a broad sector coverage
suitable for satellite applications.
BIBLIOGRAPHY
1. J. D. Kraus, Antennas, 2nd ed., New York: McGraw-Hill, 1988,
pp. 265339.
2. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
2nd ed., New York: Wiley, 1998, pp. 231239.
3. W. W. Hansen and J. R. Woodyard, A new principle in directional
antenna design, IRE Proc., 26: 333345, 1938.
4. C. A. Balanis, Antenna Theory: Analysis and Design, 2nd ed., New
York: Wiley, 1997, pp. 271276.
5. H. Nakano, Helical and Spiral Antennas: A Numerical Approach,
New York: John Wiley and Sons, 1987, pp. 123195.
6. R. S. Elliott, Antenna Theory and Design, Englewood Cliffs, NJ:
Prentice-Hall, 1981, pp. 7178.
7. H. E. King and J. L. Wong, Characteristics of 1 to 8 wavelength
uniform helical antennas, IEEE Trans. Antennas Propagat.,
AP28: 291296, 1980.
8. E. A. Wolf, Antenna Analysis, New York: Wiley, 1967, pp.
437444.
9. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, 2nd
ed. New York: McGraw-Hill, 1984, pp. 13-113-23.
10. J. L. Wong and H. E. King, Broadband quasi-taper helical antennas. IEEE Trans. Antennas Propagat., AP-27: 7278, 1979.
11. L. O. Krause, Sidefire helix UHF-TV transmitting antenna, Electronics, 24: 107109, Aug. 1951.
12. J. D. Kraus, Radio Astronomy, 2nd ed., Powell, OH: CygnusQuasar, 1986.
13. S. Sensiper, Electromagnetic wave propagation on helical structures, Proc. IRE, 43: 149161, 1955; also Ph.D. Thesis, M.I.T.,
1951.
14. P. W. Klock, A Study of Wave Propagation of Helices, Ph.D. Thesis,
University of Illinois, Urbana-Champaign, 1963.
15. C. Gerst and R. A. Worden, Helix antennas take turn for better,
Electronics, 39: 100110, Aug. 1966.
16. A. T. Adams et al., The quadrifilar helix antenna, IEEE Trans.
Antennas Propagat., AP-22: 173178, 1974.
17. A. T. Adams and C. Lumjiak, Optimization of the quadrifilar helix antenna, IEEE Trans. Antennas Propagat., AP-19: 547548,
1971.
18. W. T. Patton, The Backfire Helical Antenna, Ph.D. Thesis, University of Illinois, Urbana-Champaign, 1963.
19. C. C. Kilgus, Shaped-conical radiation pattern performance of the
backfire quadrifilar helix, IEEE Trans. Antennas Propagat., AP23: 392397, 1975.
20. J. M. Tranquilla and S. R. Best, A study of the quadrifilar helix
antenna for global positioning system (GPS) applications, IEEE
Trans. Antennas Propagat., AP-38: 15451550, 1990.
21. N. Padros et al., Comparative study of high-performance GPS receiving antenna designs. IEEE Trans. Antennas Propagat., AP45: 698706, 1997.
ARLON T. ADAMS
JAY K. LEE
Syracuse University
HORN ANTENNAS
A horn antenna is a length of conducting tube that is flared at one end and used for the transmission and
reception of electromagnetic waves. For an efficient transition between guided and radiated waves the horn
dimensions must be comparable to the wavelength. Consequently, horns are used mostly at centimeter and
millimeter wavelengths. At lower or higher frequencies they are inconveniently large or small, respectively.
They are most popular at microwave frequencies (3 GHz to 30 GHz) and higher frequencies (30 GHz to 300
GHz), as antennas of moderate directivity or as feeds for reflectors or elements of arrays.
Since acoustic horns have been in use since prehistoric times, the design of horns as musical instruments
was a highly developed art well before the appearance of the first electromagnetic horns. This occurred shortly
after Hertz first demonstrated the existence of electromagnetic waves in 1888. Experimenters placed their spark
gap sources in hollow copper tubes [Figs. 1(a) and 5(a)]. These tubes acted as high-pass filters for microwave and
millimeter-wave radiation from the open end. In London in 1897, Chunder Bose used rectangular conducting
tubes with collecting funnels, or pyramidal horns [Fig. 1(d)], in his demonstrations at 5 mm and 25 mm
wavelengths [see Ramsay (1)]. Thus the electromagnetic horn antenna was introduced, but this early beginning
of microwave invention closed with Marconis demonstration that longer wavelengths could be received at
greater distances. Horns were too large to be practical at those wavelengths, and it was almost 40 years before
microwave horns reappeared with the need for directive antennas for communications and radar. Horns alone
were often not sufficiently directive; but combined in an array or with a lens [Fig. 4(a)], or more often a parabolic
reflector [Figs. 4(b) and 4(c)], highly directive antenna beams are obtained.
HORN ANTENNAS
Fig. 1. (a) Open-ended rectangular waveguide. (b) E-plane sectoral horn. (c) H-plane sectoral horn. (d) Pyramidal horn.
when x cos . Thus the phase variation in radians across the aperture for small flare angles is approximately kx2 /(2), where k = 2/ is the propagation constant. This quadratic phase variation increases
with increasing flare angle, thus reducing directivity increase due to the enlarged aperture dimension. It is
HORN ANTENNAS
which is the approximate difference in wavelengths between the distance from the apex to the edge (x = d/2)
and the center (x = 0) of the aperture. The radiation patterns of Figs. 3(a) and 3(b) [from Love (2)] show the
effect of increasing s on the E- and H-plane radiation patterns of sectoral and pyramidal horns. The main beam
is broadened, the pattern nulls are filled, and the sidelobe levels raised over those for an in-phase aperture
field (s = 0). With large flare angles, radiation from the extremities of the aperture can be so out of phase with
that from the center that the horn directivity decreases with increasing aperture width.
The adverse effects of the flare can be compensated by a lens in the aperture [Fig. 4(a)]; but because that
adds to the weight and cost and because bandwidth limitations are introduced by matching the lens surfaces
to reduce reflections, it is not usually done except in some millimeter wave applications. Instead a combination
of aperture width and flare length in wavelengths is chosen which provides maximum axial directivity or
minimum beamwidth. This is an optimum horn design. To achieve higher directivity or narrower beamwidth
for a given aperture width, a longer horn is required.
Sectoral horns [Figs. 1(b) and (c)] are rectangular waveguides flared in one dimension only. The incident
waveguide mode becomes a radial cylindrical mode in the flared region of the horn. Since radiation pattern
beamwidths are inversely proportional to aperture dimensions in wavelengths, sectoral horns have beams
which are narrow in the plane containing the broad dimension. Such fan-shaped beams may be useful for
illuminating elongated parabolic reflectors or parabolic cylinder reflectors.
A pyramidal horn [Fig. 1(d)] is flared in both waveguide dimensions and thus is more adaptable both
as a reflector feed and on its own. The forward radiation pattern may be calculated quite accurately from
Kirchhoff diffraction theory for all but small horns. The TE01 rectangular waveguide mode yields an aperture
field uniform in one dimension (in the E plane) and cosinusoidal in the other (the H plane). A comparison of
Figs. 3(a) and 3(b) shows that this results in higher sidelobes in the E plane and, for a square aperture, a
narrower beam. Pyramidal horns are relatively easily constructed, and for all but small horns their axial gain
can be predicted accurately. Consequently they are used as gain standards at microwave frequencies; that is,
HORN ANTENNAS
Fig. 3. Universal radiation patterns of sectoral and pyramidal horns flared in the (a) E plane and (b) H plane. The
parameter s equals b2 /8E in (a) and a2 /8H in (b), and 2s/ is the maximum phase difference between the fields at the
c 1984 McGraw-Hill, Inc., from Love (2)].
center and the edge of the aperture [
they are used to experimentally establish the gain of other microwave antennas by comparing their response
with the same illuminating field.
Most of the preceding remarks on open-ended rectangular waveguides and pyramidal horns apply also
to open-ended circular waveguides and conical horns [Figs. 5(a) and (b)]. For propagation of the lowest-order
mode (TE11 ), only in a circular waveguide the interior diameter must be 0.59 < a < 0.77. This mode has a
uniform aperture field in the E plane and an approximately cosinusoidal distribution in the orthogonal H
plane. This appears, modified by a quadratic phase variation introduced by the flare, in the aperture field of the
horn. Consequently the E-plane radiation pattern of the horn is narrower, but with higher sidelobes than the
H-plane pattern and the radiated beam is elliptical in cross section. In addition, cross-polarized fields appear
in pattern lobes outside the principal planes.
HORN ANTENNAS
Fig. 4. (a) Horn aperture field phase correction by a lens. (b) Parabolic reflector fed by a horn. (c) Horn reflector antenna.
f is the focal length of the lens or reflector.
to originate, must be placed at the focus of the reflector [Fig. 4(b)]. This phase center is within the horn on the
horn axis and depends on the flare angle and aperture distribution. For both rectangular and conical horns
the position of the phase center is not the same in the E and H planes, or planes containing the electric and
magnetic field vectors, respectively. A phase center can be calculated from the average of the positions of the
HORN ANTENNAS
Fig. 5. (a) Open-ended circular waveguide. (b) Conical horn. (c) Corrugated horn. (d) Circular waveguide with corrugated
flange.
E- and H-plane phase centers, or it can be determined from the position of the feed which maximizes the gain
of the reflector antenna.
For efficient reflector aperture illumination the feed horn radiation pattern should approximately match
the shape of the reflectors aperture, and illuminate it essentially uniformly and with minimal spillover, or
radiation missing the reflector. Pyramidal horns may seem suitable for rectangular apertures because their
HORN ANTENNAS
beams are rectangular in cross section, and conical horns may seem a natural choice for a circular aperture;
but efficient aperture illumination is not obtained in either case, because their principal plane patterns differ.
Both horns have high E-plane pattern sidelobes and low H-plane sidelobes. A dual (TE11 /TM11 ) mode conical
horn provides equal E- and H-plane beamwidths and equally low sidelobes and is an efficient feed for a
circular aperture over a narrow frequency band [Love (3), p. 195]. (Reference 3 contains reprints of most earlier
significant articles on horn antennas.) A broadband solution achieves an axisymmetric beam with annular
corrugations on the interior surfaces of a conical horn [Fig. 5(c)] [Love (3), pp. 248, 277]. These produce a
horn aperture field distribution which is approximately cosinusoidal across the conical horn aperture in all
directions and hence an axisymmetric radiation pattern with low sidelobes. Such corrugations in the E-plane
interior walls only of a pyramidal horn will produce a nearly cosinusoidal E-plane aperture distribution, and
consequently similar E-plane and H-plane radiation patterns for a square horn aperture.
A feed for a small circular aperture reflector which is more easily constructed than a corrugated conical
horn, but with a less axisymmetric radiation pattern, is an open-ended circular waveguide ringed by a recessed
disc of approximately quarter-wavelength-deep corrugations [Fig. 5(d)]. These corrugations suppress back radiation from the feed and thus improve the aperture illumination over that of a simple open circular waveguide
[Love (3), pp. 181, 226]. Combined with dual mode excitation, this arrangement provides an efficient field for
a paraboloidal reflector.
Here
and
HORN ANTENNAS
Fig. 6. Coordinates for radiation from (a) rectangular and (b) circular apertures.
is a vector defining the angular behavior of the radiation polarization for an aperture in space. For an aperture
in a conducting plane, it is more accurate to use
which, since it is based on the aperture plane electric fields only, fully satisfies the assumption of a vanishing tangential field in the aperture plane outside the aperture. Consequently, radiation fields of open-ended
waveguides and small horns can be calculated accurately from Eq. (3) with Eq. (6) if they are mounted in a
conducting plane. Clearly, Eqs. (5) and (6) differ significantly only for large angles off the beam axis.
If the aperture field is separable in the aperture coordinatesthat is, in Eq. (3), Ex (x, y, 0) = E0 E1 (x)E2 (y),
where E1 (x) and E2 (y) are field distributions normalized to E0 then the double integral is the product of two
single integrals:
where
HORN ANTENNAS
Open-Ended Waveguides
Rectangular Waveguides. With the TE01 waveguide mode the aperture field
in Eq. (7) yields the following for Eqs. (8) and (9):
This defines the radiation pattern in the forward hemisphere /2 < < /2, 0 < < 2. If the aperture is in
space, then Eq. (5) is used for A(r, , ), but this is not an accurate solution since the aperture dimensions are
not large. Rectangular waveguides mounted in conducting planes use Eq. (6) for A(r, , ) in Eq. (7), which then
accurately provides the far field. The pattern has a single broad lobe with no sidelobes. For large apertures,
plots of the normalized E-plane ( = 0) and H-plane ( = /2) patterns of Eq. (7) appear in Figs. 3(a) and 3(b)
for those of a horn with no flare (s = 0), but without the factor (1 + cos )/2 from Eq. (5) or cos from Eq. (6).
Circular Waveguides. The dominant TE11 mode field in circular waveguide produces an aperture field
distribution which in the aperture coordinates , of Fig. 6(b) is
where J 1 is the Bessel function of the first kind and order, J 1 is its derivative with respect to its argument kc ,
and kc a/2 = 1.841 is the first root of J 1 . E0 is the electric field at the aperture center ( = 0). Since Eq. (13) is
not linearly polarized, its use in Eq. (3) provides only part of the total radiated far field. This total field is
in which k = k sin .
In the E and H planes ( = 0 and /2) the cross-polarized fields cancel and the patterns shown in Fig. 14(a)
are similar to those of Eqs. (11) and (12), respectively, but with slightly broader beams and lower sidelobes for
the same aperture dimensions. As with rectangular waveguides, open-ended circular waveguide apertures are
10
HORN ANTENNAS
insufficiently large for Eq. (14) to represent all the radiated fields accurately. In the principal planes ( = 0,
/2) it can give a reasonable approximation for the copolarized fields but fails to accurately represent the crosspolarized field patterns in = /4. This is evident from a comparison of numerical results from approximate
and exact solutions [Collin (4), p. 233].
HORN ANTENNAS
11
which are tabulated and for which computer subroutines are available. For example,
with
Figure 3(a) shows plots of the magnitude of Eq. (22) for various values of the E-plane flare parameter s = b2 /8E ,
while Fig. 3(b) shows corresponding plots of |I2 (k sin )/I2 (0)| for the H-plane flare parameter s = a2 /8H . For
no flare (s = 0) the patterns are those of a large open-ended rectangular waveguide supporting only the TE01
mode. The effect of the flare is to broaden the main beam, raise the sidelobes, and fill pattern nulls. For larger
values of s there is enhanced pattern beam broadening and eventually a splitting of the main beam on its axis.
These curves also represent the radiation patterns of the E- and H-plane sectoral horns of Figs. 1(b)
and 1(c). For an E-plane sectoral horn (H ) the E-plane pattern is given by Eq. (19) and the H-plane
pattern is given approximately by Eq. (12). For an H-plane sectoral horn (E ) the E-plane pattern is given
approximately by Eq. (11) and the H-plane pattern is given by Eq. (20).
In comparing Figs. 3(a) and 3(b) it is evident that E-plane beamwidths of a square aperture are narrower
than H-plane beamwidths. For horns of moderate flare angle and optimum horns the E-plane half-power
beamwidth is 0.89 /b radians and the H-plane half-power beamwidth 1.22 /a radians. E-plane patterns have
minimum sidelobes of 13.3 dB below peak power while H-plane pattern minimum sidelobes levels are 23.1
dB.
The universal patterns of Figs. 3(a) and (b) can also be used to predict the approximate near-field radiation
patterns of horns by including the quadratic phase error which is a first-order effect of finite range r. This is
done by including
in Eq. (15). Then the near-field principal plane patterns of a pyramidal horn are given by Eqs. (17) and (18)
with E , H replaced by
and
12
HORN ANTENNAS
These near-field effects are analogous to decreasing the flare length of a horn with a fixed aperture width. The
main beam broadens, nulls are filled in, and sidelobes rise.
Limitations and Extensions. Results from Eq. (16) do not apply to small horns and are limited to the
forward direction ( < 90 ). They are most accurate on and around the beam axis ( = 0), becoming progressively
less accurate as increases. The simplest method for extending the analysis is by the uniform geometrical theory
of diffraction [e.g., Love (3), p. 66], which provides the edge-diffracted fields in the lateral and rear directions
which receive no direct illumination from the aperture. Only the edges normal to the plane of the pattern
contribute significantly to the E-plane pattern but the rear H-plane pattern requires contributions from all
four aperture edges and so is difficult to calculate this way.
While the geometry of the pyramidal horn defies rigorous analysis, numerical methods have been used
with some success for open waveguides and small horns. For larger horns, this approach becomes computationally intensive, but some results from Liu et al. (5) are shown in Fig. 7 and compared with measurements and
approximate computations. Their numerical computations, along with measurements of the aperture fields by
Nye and Liang (6) show that higher-order modes need to be added to the dominant mode field of Eq. (15) and
that the parabolic phase approximation of Eq. (1) improves as the aperture size increases.
Gain. Pyramidal horns are used as gain standards at microwave frequencies because they can be
accurately constructed and their axial directive gain reliably predicted from a relatively simple formula. The
ratio of axial far-field power density to the average radiated power density from Eq. (16) yields
where G0 = 32 ab/(2 ) is the gain of an in-phase uniform and cosinusoidal aperture distribution. The reduction
of this gain due to the phase variation introduced by the E-plane flare of the horn is
where the Fresnel integrals and their argument are defined by Eqs. (21) and (23). Similarly the gain reduction
factor due to the H-plane flare of the horn is
where
A plot of RE and RH in decibels as a function of the parameter 2d2 /, where d is the appropriate aperture
dimension b or a and is the slant length E or H , respectively, is shown in Fig. 8. Calculation of the gain
from Eq. (28) is accurate to about 0.1 dB for 22 dB standard gain pyramidal hornsthat is, optimum horns
with dimensions of at least 5. For 18 dB gain horns the accuracy is about 0.2 dB and for 15 dB horns
0.5 dB. Since optimum gain pyramidal horns have an aperture efficiency of approximately 50%, the gain is
approximately
HORN ANTENNAS
13
Fig. 7. Calculated and measured (a) E-plane and (b) H-plane radiation patterns of a pyramidal horn of dimensions
c 1993 IEEE, from Liu et al. (6)].
a = 4.12, b = 3.06, E = 10.52, and H = 9.70 [
For an E-plane sectoral horn H and RH (v, w) 1 the axial gain is then GE = G0 RE (u), an inaccurate
formula because aperture dimension a is less than a wavelength. A result which includes the fact that aperture
electric and magnetic fields are not related by free-space conditions and that interaction occurs across the
narrow aperture of the horn is
14
HORN ANTENNAS
Fig. 8. E- and H-plane flare and near field gain reduction factors RE and RH of pyramidal and sectoral horns in decibels
c 1981 IEE, from Jull (11)].
[
where
and
is the guide wavelength. The accuracy of Eq. (33) is comparable to that of Eq. (28) for the horns of similar b
dimension.
The gain of an H-plane sectoral horn, obtained by letting E so that RE (u) 1, is GH = G0 RH (v, w).
It probably is reasonably accurate, but there appears to be no experimental evidence available to verify it.
The near-field gain of pyramidal and sectoral horns can be calculated from the above expressions by
replacing E and H by Eqs. (26) and (27), respectively.
Conical Horns
The aperture field of a conical horn fed by a circular waveguide supporting the TE11 mode is approximately
HORN ANTENNAS
15
Fig. 9. Copolar and cross-polar radiation patterns for a conical horn with dimensions a = 4 and = 23. E plane,
c 1994 IEE, from Olver et al. (7)].
H plane, cross-polarization [
where E( , ) is given by Eq. (13) and is the slant length of the horn. Numerical calculation of the radiation
patterns is necessary. In the example of Fig. 9 [Olver et al. (7)] with a flare angle = 5 and aperture width
a = 4, the E-plane ( = 0) pattern is narrower than the H-plane ( = /2) pattern as in square rectangular
horns. The cross-polar ( = /4) radiation pattern peak level is 18.7 dB relative to the copolar pattern peak
levels, a level typical of conical horn apertures larger than about 2. Smaller conical horns can have more
axisymmetric patterns. E- and H-plane patterns have equal beamwidths for an aperture diameter a = 0.96,
and cross-polarized fields cancel for a = 1.15. This makes small conical horns efficient as reflector feeds and
as array elements with high polarization purity.
Larger conical horns are similar to rectangular horns in their lack of axial pattern symmetry. Optimum
gain conical horns have an aperture efficiency of about 54% and half-power beamwidths in the E and H planes
of 1.05/a and 1.22/a radians, respectively, for aperture diameters of more than a few wavelengths.
16
HORN ANTENNAS
c 1984 McGraw-Hill,
Fig. 10. Excitation of axisymmetric linearly polarized aperture fields in a stepped conical horn [
Inc. from Love (2)].
this is done by exciting the TM11 mode, which propagates for waveguide diameters a > 1.22, in addition to
the TE11 mode, which propagates for a > 0.59. The electric field configuration of these modes in a waveguide
cross section is shown in Figs. 10(a) and (b). Added in phase and in the right proportion, cross-polarized and
aperture perimeter fields cancel, while the copolar fields around the aperture centre add, yielding the aperture
HORN ANTENNAS
17
field configuration of Fig. 10(c). These mixed mode fields are linearly polarized and tapered approximately
cosinusoidally radially across the aperture. This yields the essentially linearly polarized and axisymmetric
radiation patterns desired.
Partial conversion of TE11 to TM11 fields can be effected by a step discontinuity in the circular waveguide
feed, as in Fig. 10(d), or by a circular iris or dielectric ring in the horn. The TM11 /TE11 amplitude ratio depends
on the ratio of waveguide diameters, and the relative phase of the modes depends on the length of largerdiameter circular waveguide and the horn. This limits the bandwidth of the horn to about 5%. A multimode
square-pyramidal horn has similar low sidelobe level E- and H-plane radiation patterns due to an essentially
cosinusoidal aperture distribution in the E-plane as well as the H-plane [Love (2)]. This can be achieved by
excitation of a hybrid TE21 /TM21 mode by an E-plane step discontinuity or by changes in the E-plane flare.
Corrugated horns have aperture fields similar to those of Fig. 10(c) and consequently similar radiation
patterns, but without the frequency bandwidth limitations of the above horn. This is achieved by introducing
annular corrugations in the interior walls of a conical horn. There must be sufficient corrugations per wavelength (at least three) that the annular electric field E is essentially zero on the interior walls. The corrugations
make H also vanish. This requires corrugation depths such that short circuits at the bottom of the grooves
appear as open circuits at the top, suppressing axial current flow on the interior walls of the horn. This groove
depth is /4 on a plane corrugated surface or a curved surface of large radius. For a curved surface of smaller
radius, such as near the throat of the horn, the slot depths need to be increased; for example, for a surface
radius of 2, the depth required is 0.3. Usually slots are normal to the conical surface in wide flare horns but
are often perpendicular to the horn axis with small flares. To provide a gradual transition from the TE11 mode
in the waveguide to a hybrid HE11 mode in the aperture, the depth of the first corrugation in the throat should
be about 0.5 so that the surface there resembles that of a conducting cone interior. Propagation in corrugated
conical horns can be accurately calculated numerically by mode-matching techniques. The aperture field is
approximately
where kc a/2 is 2.405, the first zero of the zero-order Bessel function J 0 , is the slant length of the horn, and A is
a constant. This aperture field is similar to that of Fig. 10(c), and the resulting E and H patterns are similarly
equal down to about 25 dB. Some universal patterns are shown in Fig. 11. Cross-polarization fields are also
about 30 dB from the axial values, but are now over a bandwidth of 2:1 or more.
Broadband axisymmetric patterns with low cross-polarization make corrugated horns particularly attractive as feeds for reflectors. Low cross-polarization allows the use of dual polarization to double the capacity
of the system. Another notable feature for this application is that the position of the E- and H-plane pattern
phase centers coincide. Figure 12 shows the distance of the phase center from the horn apex, divided by the
slant length, of small-flare-angle conical [Milligan (8)] and corrugated [Thomas (9)]] horns for values of the
phase parameter s given by Eq. (2). For a conical horn the E-plane phase center is significantly further from
the aperture than the H-plane phase center. Thus if a conical horn is used to feed a parabolic reflector, the best
location for the feed is approximately midway between the E- and H-plane phase centers. With a corrugated
horn, such a compromise is not required so it is inherently more efficient.
Corrugated horns may have wide flare angles, and their aperture size for optimum gain decreases correspondingly. For example, with a semiflare angle of 20 the optimum aperture diameter is about 8, whereas
for a semiflare angle of 70 it is 2. Wide-flare corrugated horns are sometimes called scalar horns because of
their low cross-polarization levels.
18
HORN ANTENNAS
c 1984
Fig. 11. Universal patterns of small-flare-angle corrugated horns as a function of the parameter s = a2 /8 [
McGraw-Hill, Inc. from Love (2)].
Prole Horns
Most corrugated horns are conical with a constant flare angle. Figure 13 shows a profile conical horn in which
the flare angle varies along its length. This arrangement provides a horn shorter than a conical corrugated
horn of similar beamwidth, with a better impedance match due to the curved profile at the throat and an
essentially in-phase aperture field distribution due to the profile at the aperture. Consequently the aperture
efficiency is higher than that of conical corrugated horns. The phase center of the horn is near the aperture
center and remains nearly fixed over a wide frequency band. Radiation patterns of a profile horn similar to
that of Fig. 13 are shown in Fig. 14 [Gonzalo et al. (10)]. The patterns are similar to those of a Gaussian beam,
such as is radiated from the end of an optical fiber supporting the HE11 mode. The performance of this small
horn as a feed seems close to ideal, but larger profile horns may exhibit higher sidelobe levels due to excitation
of the HE12 mode at the aperture.
Horn Impedance
Antennas must be well matched to their transmission lines to ensure a low level of reflection in microwave
communication systems. The impedance behavior of a horn depends on the mismatch at the waveguide/horn
junction and at its aperture. For an E-plane sectoral horn, reflections from these discontinuities are comparable
in magnitude; and since they interfere, the total reflection coefficient oscillates with frequency and the input
voltage standing wave ratio (VSWR) may vary from 1.05 at high frequencies to 1.5 at the lowest frequency. With
HORN ANTENNAS
19
Fig. 12. Normalized distance of the phase center from the apex of conical ( E plane, H plane) and corrugated ()
horns [Data from Milligan (8) and Thomas (9)].
H-plane sectoral horns, aperture reflection is much stronger than junction reflection so their VSWR increases
almost monotonically with decreasing frequency. An inductive iris in the waveguide near the E-plane horn
junction can match its discontinuity. A capacitive iris may be similarly used for an H-plane sectoral horn.
Aperture reflections in these horns may be matched with dielectric covers.
Pyramidal horns of sufficient size and optimum design tend to be inherently well-matched to their waveguide feeds because the E- and H-plane aperture and flare discontinuities partially cancel. For example, a 22
dB gain horn has a VSWR of about 1.04 while an 18 dB horn has a VSWR of less than 1.1.
Conical horns fed by circular waveguides supporting the dominant TE11 mode have an impedance behavior
similar to that of pyramidal horns of comparable size fed by rectangular waveguides. The waveguide/horn
discontinuities of both horns may be matched by an iris placed in the waveguide near the junction. A broader
20
HORN ANTENNAS
Fig. 14. (a) Far-field radiation patterns of TE11 mode and (b) radiation patterns of a profile corrugated horn of aperture
c 1997, IEEE, from
a = 15.8 mm and length L = 26.7 mm at 30 GHz. E plane,H plane cross-polarization [
Gonzalo et al. (10)].
bandwidth match is provided by a curved transition between the interior walls of the waveguide and the horn.
Broadband reduction of aperture reflection may be similarly reduced by a curved surface of a few wavelengths
radius. Such aperture-matched horns also have lower sidelobe levels and less back radiation in their E-plane
patterns than do conventional pyramidal and conical horns. Their H-plane flare patterns are affected little by
such aperture matching because the electric field vanishes at the relevant edges.
For dual-mode and corrugated horns there are also negligible fields at the aperture edges and hence little
diffraction there. Corrugated horns with initial groove depths near the throat of about a half-wavelength and
which gradually decrease to a quarter-wavelength near the aperture, as in Fig. 13, are well-matched at both
throat and aperture. For most well-designed corrugated horns a VSWR of less than 1.25 is possible over a
HORN ANTENNAS
21
frequency range of about 1.5:1. Dual-mode horns using a step discontinuity as in Fig. 10(d) may have a VSWR
of 1.2 to 1.4. If an iris is required for a match, the frequency bandwidth will of course be limited. Conical and
pyramidal horns using flare-angle changes to generate the higher-order modes can have VSWRs less than 1.03
and require no matching devices.
BIBLIOGRAPHY
1. J. F. Ramsay, Microwave antenna and waveguide techniques before 1900, Proc. IRE, 46: 405415, 1958.
2. A. W. Love, Horn antennas, in R. C. Johnson and H. Jasik (eds.), Antenna Engineering Handbook, 2nd ed., New York:
McGraw-Hill, 1984, chap. 15.
3. A. W. Love (ed.), Electromagnetic Horn Antennas, Piscataway, NJ: IEEE Press, 1976.
4. R. E. Collin, Antennas and Radiowave Propagation, New York: McGraw-Hill, 1985.
5. K. Liu et al., Analysis of pyramidal horn antennas using moment methods, IEEE Trans. Antennas Propag., 41: 1379
1389, 1993.
6. J. F. Nye, W. Liang, Theory and measurement of the field of a pyramidal horn, IEEE Trans. Antennas Propag., 44:
14881498, 1996.
7. A. D. Olver et al., Microwave Horns and Feeds, vol. 39, London: IEE Electromagnetic Waves Series, 1994.
8. T. Milligan, Modern Antenna Design, New York: McGraw-Hill, 1985, chap. 7.
9. B. M. Thomas, Design of corrugated horns, IEEE Trans. Antennas Propag., 26: 367372, 1978.
10. R. Gonzalo, J. Teniente, C. del Rio, Very short and efficient feeder design from monomode waveguide, IEEE Antennas
Propag. Soc. Int. Symp. Dig., Montreal, 1997, pp. 468470.
11. E. V. Jull, Aperture Antennas and Diffraction Theory, vol. 10, London: IEE Electromagnetic Waves Series, 1981.
EDWARD V. JULL
University of British Columbia
LEAKY-WAVE ANTENNAS
Leaky-wave antennas (LWAs) constitute a type of radiators whose behavior can be described by an electromagnetic wave (leaky wave) that propagates in guiding structures that do not completely confine the field, thus
allowing a continuous loss of power to the external environment (leakage).
According to IEEE Standard 145-1983, a leaky-wave antenna is an antenna that couples power in small
increments per unit length either continuously or discretely, from a traveling wave structure to free space.
LEAKY-WAVE ANTENNAS
Fig. 1. Basic structures of LWAs: (a) uniform LWAs: geometry representing a partially open metallic waveguide; (b)
periodic LWAs: geometry representing a strip-loaded dielectric-rod waveguide; (c) topology of LWA arrays.
forward and in the backward quadrant. However, for various reasons, there are limitations on such scanning
ranges for both types of structures (as shown in a subsequent section).
When a pencil beam is to be used for two-dimensional (2-D) scanning, both in elevation and cross planes
(zenith and azimuth), a phased array of juxtaposed LWAs may be employed, thus enlarging the equivalent
aperture transversely (11,12) [see the reference example of Fig. 1(c)]. LWA arrays are therefore constituted
by a linear configuration of sources (i.e., 1-D elements), instead of the planar ones of standard arrays (i.e.,
2-D elements). For LWA arrays pointed-beam scanning is achievable by varying both the frequency for the
elevation plane and the phase shift for the cross plane.
Since LWAs are derived from partially open waveguides, they present a number of distinctive features
as radiators: large power-handling capability, particularly for structures utilizing closed metallic waveguides;
reduction of bulk, due to the usually small profiles in the cross sections; capability of designing a wide variety
of aperture distributions and consequent flexibility for the beam shaping; possible use as wide-band radiators,
allowing large angular scanning by varying the frequency (instead of using mechanical or other electronic
means); achievement of very narrow beams with good polarization purity; simplicity of feeding and economy
for 2-D scannable pencil-beam arrays (multiple linear phase shifters instead of one planar one); etc.
LWAs are mainly used in the microwave and millimeter-wave region, though investigations have been
performed also for both lower- and higher-frequency applications. The first studies on LWAs were presented
during the forties, basically for aerospace applications (radar, etc.); since then, a very wide range of solutions
LEAKY-WAVE ANTENNAS
for LWAs has been proposed in response to changing requirements and constraints. The applicability of this
type of antennas has also been widened, including various problems of traffic control, remote sensing, wireless
communications, and so forth (13,14).
Operation Principles
Existence and excitation of leaky waves in open guides. A leaky wave has a complex longitudinal
wave-number kz that can be derived by solving, as a function of the physical parameters (frequency and
geometry of an open waveguiding structure), the characteristic equation (or dispersion relation), which is of
the general type
where k0 = (0 0 ) 2 is the vacuum wave-number. [The eigensolutions of Eq. (1) can alternatively be viewed as
the pole singularities of the spectral Greens function of the structure, since the expression D(kz ,k0 ) is generally
present in its denominator (5,6,7,8,9,10).]
As is well known, for lossless closed waveguides the dispersion relation in Eq. (1) generally presents an
infinite discrete set of eigensolutions giving the guided modes (proper modes) which individually satisfy all
the relevant boundary conditions. Any field excited by a source in a closed guide can be expanded in terms
of the complete set of the infinite discrete eigensolutions of Eq. (1). In conventional guides, the longitudinal
1
wave numbers kz are either real [propagating waves above their cutoff, with kz = z < k = k0 (r ) 2 ; for air-filled
structures, one has z < k0 ] or imaginary (attenuating waves below their cutoff, with kz = jz ). (We recall that
in reciprocal structures such eigensolutions exist in pairs: kz = kz , describing waves traveling in opposite z
directions.)
In lossless open waveguides (e.g., dielectric guides), instead, only a finite number of propagating modes can
exist as eigensolutions of Eq. (1) satisfying all the boundary conditions (particularly, the radiation condition):
these are the so-called bound surface waves (each one exists only above its cutoff, with kz 2 > k0 2 , or kz = z >
k0 ). In addition to this, for a complete representation of the field, which is no longer confined in a closed section,
a continuous spectrum of modes must be introduced: it describes the radiated field as an integral over a set of
plane waves having a continuous range of wave numbers (e.g., such that < kz 2 < k0 2 , or 0 kz = z k0 and
j < kz = jz < j0) (5,6,7,8,9,10). Any field excited by a source in an open guide can therefore be expanded
in a spectral representation, that is, in terms of a finite set of proper modes and an integral contribution of the
continuous spectrum. (Again, in reciprocal structures the solutions exist in pairs: kz .)
On the other side, it is seen that the characteristic equation in Eq. (1) for open guides presents additional
discrete solutions that are nonspectral in that they correspond to fields that violate the radiation condition (they
attenuate along the propagation direction but increase exponentially in a transverse direction away from the
structure) and are not included in the spectral representation of the field. These roots are also called improper,
since they are located on the improper sheet of the Riemann surfaces for kz (5,6,7,8,9,10).
In many practical circumstances, for describing the radiative effects of the open structures in the presence of a source, the evaluation of the field through the spectral representation (i.e., including the integral
contributions of the continuous spectrum) can be very difficult and cumbersome. It is found that the field can
generally be evaluated in an alternative and rapidly convergent way, through the so-called steepest descent
or saddle point representation (5,6,7,8,9,10): by properly modifying the integration path of the continuous
spectrum involved in the spectral representation, the relevant integral evaluation can be simplified by introducing contributions of improper eigensolutions. The advantage of such a steepest-descent representation is
that, for guiding structures designed to operate as LWAs, the radiation field can be evaluated accurately in
LEAKY-WAVE ANTENNAS
a much simpler fashion by considering just the residual contribution due to the presence of one complex and
generally improper mode, that is, the leaky wave, which can therefore be viewed as a simple recasting of the
continuous spectrum. In fact, it is seen that in practical cases the remaining part of the continuous spectrum
in the steepest-descent representation (the so-called space wave) can be made to give negligible contributions
to the description of the LWAs radiation (5,6,7,8,9,10).
The existence of a leaky mode, which describes a wave propagating with a certain amount of attenuation
through the aperture of a traveling-wave structure, is a necessary condition to ensure its influence on the
illumination function. Furthermore, the leaky wave should be dominant, in the sense that its excitation has to
be strong and therefore its amplitude should be large in comparison with those of other possible guided modes
and of the space wave. [A deep discussion of the physical nature of the complex waves and its influence on the
evaluation of the field in open structures cannot be given within the limits of this article, but many detailed
studies of it can be found in the literature (1,2,3,4,5,6,7,8,9,10).]
From another point of view it can be seen that, when properly excited by a source at a finite section, a leaky
wave, even though improper, assumes its physical validity within an angular sector close to the equivalent
aperture of the open guiding structure, and the corresponding field distribution is able to furnish a fundamental
contribution to the evaluation of the near field (5,6,7,8,9,10). Since the far field is expressed as a simple Fourier
transform of the field at the aperture, a leaky wave can furnish a highly convergent and efficient evaluation of
the radiation of LWAs, as an extremely advantageous alternative to a continuous-spectrum evaluation.
In an open lossless structure the leaky-wave solutions that are of the type kz = z jz describe power
flowing away from the structure. Actually, in usual reciprocal structures these roots of the dispersion equation
in Eq. (1) appear in pairs of the type kz = kz . This means that, in general, both direct and reverse waves
exist, propagating in the +z and z directions (i.e., the kz =z jz and kz = z + jz solutions); as mentioned,
in standard applications the source is chosen to excite just one of the two opposite traveling waves. In addition,
complex conjugate solutions (kz = z + jz and kz = z jz ) also occur for describing leakage phenomena
related to power entering the guiding structure from the external region, because the field increases along the
propagation direction; again, in standard applications the source is placed inside the guiding structure and
leakage occurs only through one wave, which carries out power. It should be also mentioned that, in some cases
(e.g., the nonreciprocal structures, as discussed in a subsequent section), complex waves of the type kz = z
jz and its conjugate kz may exist, which describe effects of power leakage as well and also possess a spectral,
or proper, nature.
Basic characterization of leaky-wave antennas. LWAs present the advantage of a rather simple
characterization of their basic properties, with consequent straightforward approaches for their analysis and
synthesis. As mentioned, their characterization is reduced to the evaluation of a dominant complex eigensolution kz = z jz that can be supported and strongly excited in a specific open structure.
The characteristic dispersion of the real and imaginary parts of the longitudinal wave number of a leaky
wave is presented in Fig. 2 for two different types of structures; specifically, the normalized parameters z /k0 and
z /k0 are plotted against frequency f . The radiation region of LWA structures lies largely inside the frequency
range where the wave becomes fast (z /k0 < 1) and power can therefore leak out from the guiding structure
towards the outside air region in the typical form of a TEM-like mode: in fact, z /k0 < 1 is in general the
so-called condition for leakage of a complex wave that can radiate into an external air region.
The valid frequency range for LWA applications is actually where, as the frequency decreases, z /k0
diminishes monotonically from unity towards rather low values; in this region, to have an efficient directive
beam, z /k0 should assume rather small values (e.g., typically z /k0 can be of the order from about 10 1 to 10 3 ).
As frequency decreases further, a sudden rise of z /k0 is generally found, which signifies the predominance of
reactive phenomena over radiative ones, while z /k0 may present a flat zone with approximately constant low
values before showing a further steep rise as the frequency goes to zero; in these ranges, radiative effects can
no longer be represented by the leaky wave, and the structures usually cannot work well as radiators (2,7,8,12).
LEAKY-WAVE ANTENNAS
Fig. 2. Typical dispersion behavior of the leaky-mode complex wave number (normalized phase constant z /k0 and leakage
constant z /k0 versus frequency f ): (a) for an open planar waveguide (microstrip); (b) for a partially closed metallic waveguide
(slit ridge).
The case of Fig. 2(a) is representative in particular of leakage in unperturbed intrinsically open waveguiding structures (e.g., dielectric and printed lines). In this case, the leaky waves are complex roots of the same
characteristic equation D(kz ,k0 ) = 0 that furnishes the guided (or bound) modes of the structure. Each leaky
mode typically appears as a mathematical continuation of a related guided mode; at higher frequencies the
LEAKY-WAVE ANTENNAS
mode is generally slow and bound (real and proper), while at lower frequencies it becomes fast (complex and
improper) and can radiate (z /k0 < 1 and z = 0).
It is worth noting here that in open planar structures a different type of leakage can occur as well, which
is associated with surface waves (TE- or TM-like modes) propagating in the substrates (12), instead of the
standard space wave (TEM-like mode) that carries power out into the air. While the latter is able to describe
useful far-field radiation in LWA applications, the former usually describes power that leaks out transversely
in a layered structure and accounts for loss and interference effects in the planar circuits (see also a subsequent
section for comments on this phenomenon).
The case of Fig. 2(b) is representative of leakage in closed metallic waveguides (in this case, filled with
dielectric) that are perturbed with a suitable aperture to allow a certain amount of power loss to the environment. In this case the leaky waves are the complex roots of a characteristic equation that appears as a small
perturbation of that one for the closed waveguide, in such a way that the leakage mechanism also is properly characterized, having similar phase properties and the additional presence of an attenuation (leakage)
constant (see next section). In fact, from Fig. 2(b) it is seen that the behavior of z /k0 resembles that of the
closed structure until the cutoff region is approached, where reactive effects become predominant. Also, for
an open metal guide that is air-filled, the wave is always fast at any frequency (the condition for leakage is
always satisfied), while for an open metal guide that is partly filled with a dielectric the wave is fast at lower
frequencies and becomes slow at higher frequencies, where it can no longer radiate (z /k0 > 1), even in the
presence of nonzero z .
Evaluation of the leaky-wave phase and leakage constants. The evaluation of the complex eigensolutions for nonclosed waveguides depends on the physical parameters involved (frequency and geometry)
and is generally achievable with numerical methods. Among them, the transverse resonance technique (TRT)
(15,16) is one of the most efficient approaches for either rigorous or approximate (according to the antenna
topology) evaluations. It first requires the introduction of a suitable equivalent transmission-line network,
which describes the transverse geometry of the structure. Then, a numerically solvable transcendental equation in terms of transverse eigenvalues kt and of physical parameters is usually achievable by imposing a
resonance condition for the equivalent circuit. The complex eigenvalue kz is derived from the additional link
to the longitudinal problem given by a separation condition for the eigenvalues (e.g., in air, k0 2 =2 0 0 = kt 2
+ kz 2 ). Where the separation condition holds rigorously also for the variables in the transverse plane (e.g., kt 2
= kx 2 + ky 2 ), the TRT in general gives the exact characteristic equation for the geometry. Even if the problem
is not rigorously separable in the transverse plane, an equivalent characteristic equation can be achieved by
the TRT, which often gives solutions with good accuracy for many practical purposes. To this aim, the network
elements that give the electromagnetic description of the various parts of the transverse geometry (such as
radiation impedance, step discontinuities, T junctions, etc.) have to be derived in convenient forms, which are
found to be simple and usually accurate (3,4,12,17,18).
When separation of variables does not strictly hold, other numerical methods can be employed to accurately determine the complex eigensolutions for open waveguides. The most appropriate choice depends on
several factors related to the computational features of the methods, the geometry of the open structures,
etc. (15,16). Among the various possible approaches, integral-equation techniques can work particularly well.
In particular, spectral-domain approaches appear well suited for the derivation of the eigensolutions in structures of the printed type (15). Different formulations based on surface integral equations (the boundary-element
method, etc.) (16) can also furnish satisfactory numerical results in terms of efficiency, versatility, and accuracy for the analysis of quite general classes of arbitrarily shaped open waveguides, allowing the rigorous
determination of the longitudinal wave number for complex modes as well (19,20).
Interpretation of the behavior of a leaky-wave antenna. As described, LWAs are described by a
fast wave that propagates through an equivalent aperture, losing power towards free space, with a leakage
amount that is usually rather limited to allow a sufficiently directive beam. The simplest LWA geometry for this
purpose is that of a closed metallic waveguide in which a suitably small aperture is introduced longitudinally
LEAKY-WAVE ANTENNAS
in order to achieve a continuous power loss along its length, as shown in Fig. 3(a) for a rectangular guide with
a slit cut in a side wall. This structure, besides having historical importance as the first proposed LWA in
1940 (1,2,3,4), can be taken as a reference structure for explaining the basic behavior of LWAs in terms of a
waveguide description.
For such a structure, a leaky wave can be considered as excited by a standard incident mode for the closed
rectangular waveguide, that is, the dominant TE10 , which travels in the +z direction with a known phase
constant 0z for a fixed choice of the physical parameters (geometry and frequency). For a sufficiently small
geometry perturbation due to the slit, the phase constant is changed just slightly to a value represented by
z , and a low leakage rate z then occurs, which, as mentioned, accounts for the longitudinal attenuation due
to the field that is no longer confined and flows also in the outside region. The propagating field inside the
waveguide and in the proximity of its aperture is therefore described by the complex longitudinal wave number
kz = z jz , whose value depends on the physical parameters [the qualitative behavior is like that shown in
Fig. 2(b), and a quantitative description is given in this final part of this subsection].
In this case the leakage is assumed along +z (z > 0 and z > 0), and on supposing that the vertical field
variations are almost negligible (ky 0), it is easily seen, from the general separation condition for waveguides
(k0 2 = 2 0 0 = kt 2 + kz 2 kx 2 + kz 2 ), that the horizontal wave-number is also complex:
since one has x x = z z . Therefore a plane wave of inhomogeneous type exists, having a complex propagation
vector k of the type
where the phase vector is directed at an angle that describes the emission of power from the guide, and
the attenuation (leakage) vector is perpendicular to , and represents attenuation along z and amplification
along x. Consequently, the field has a spatial dependence of the type
Therefore, this plane wave travels at an angle = sin 1 (z /||) with respect to broadside, carrying
out power, and its amplitude increases transversely as expected in a leaky wave. It should be noted that
the direction angle of the leaky wave is usually expressed under the approximate form sin 1 (z /k0 ),
directly derivable from the previous relation, since in general the leakage constant is numerically negligible
with respect to the phase constant. The nature of the propagation vector is sketched in Fig. 3(b), while the
distribution of equiphase and equiamplitude surfaces with respect to the decreasing power flow along the guide
is represented in Fig. 3(c). It should be recalled that, even though the leaky wave has an improper nature, the
field generated from a source located at a finite distance along z still satisfies the radiation condition, since
the field increases transversely only in a limited sector given by angles less than the value describing the
direction of power leakage (5,6,7,8,9).
The leakage mechanism and the relationship for the beam peak direction as a function of frequency can
be interpreted in a very simple way, also referring to the description of the propagation of the dominant mode
LEAKY-WAVE ANTENNAS
Fig. 3. LWA achieved with a side-wall slit in a rectangular waveguide: (a) geometry of the structure; (b) nature of the
propagation vector of the inhomogeneous plane leaky wave (phase and attenuation vectors); (c) equiphase and equiamplitude planes of the leaky wave with the appropriate leakage along the guide; (d) interpretation of leakage in terms of
superposition of homogeneous plane waves at an angle with respect to the longitudinal direction (closed and partially open
structures); (e) equivalent transverse-resonance network, resonance conditions, and network parameters for the numerical
evaluation of the leaky-wave complex wave numbers as a function of the physical parameters involved.
of our structure in terms of superposition of two uniform plane waves incident at an angle on the lateral walls
of the guide, as represented in Fig. 3(d). With the introduction of a small aperture in one wall, at each reflection
a certain amount of power (related to the aperture dimension and to the other parameters involved through z )
leaves the guide at an angle sin 1 (z /k0 ). The pointing angle is therefore near broadside for frequencies
close to the cutoff of the closed guide and tends towards endfire as frequency increases.
A quantitative description of this LWA is easily achieved with a simple analysis of the complex eigenvalue,
derivable through a modification of the dominant mode by employing a TRT (3,4,21). To this end, one requires
LEAKY-WAVE ANTENNAS
a characterization of the slit aperture in the side wall as a circuit element in the equivalent transmission
line. For the evaluation of such discontinuities a great deal of work has been done in the past decades, mainly
through variational methods (4,12,15,16,17,21). The description of the radiative and reactive effects of the slit
in the side wall of the rectangular guide can be represented by a lumped element (e.g., an admittance Y R = GR
+ jBR ) as a function of geometry and frequency. The transverse network and the relationships that allow the
derivation of the characteristic equation having complex eigensolutions are briefly reported in Fig. 3(e). The
relevant solution in the complex plane for the perturbed dominant mode describes the leaky-wave behavior.
Design Techniques
Basic radiation features. The basic design principles of LWAs are generally derivable from the knowledge of the desired beam width and of the pointing direction. In LWAs these quantities can be linked in a
straightforward way to the complex longitudinal wave-number.
The direction M of the beam maximum, in fact, is mainly related to the phase constant for the chosen frequency. It has been seen that the value of the normalized phase constant z /k0 predicts with a good
approximation the direction of the maximum radiation M according to the simple relationship
Since z has dispersive behavior, as is typical of waveguiding structures, scanning of the beam is achieved
by changing the frequency: typically, as the frequency is increased from the cutoff, the pointing angle varies
from around broadside (M = 0 ), towards forward endfire (M = 90 ). (Comments on the limitations of the
scanning capability of LWAs will be presented in the next subsection.)
As regards the beam width, we recall that the leakage constant z expresses the rate of power loss along
the line due to the radiation, thus influencing primarily the effective dimension of the equivalent aperture for
the line source. In fact, the more z increases, the more the actual illumination length decreases (and the less
the beam width is focused).
A basic relation between the leakage constant and the antenna length L derives from the specification
of the radiation efficiency , expressible in LWAs as = [P(0) P(L)]/P(0), where P(0) is the input power
delivered to the structure and P(L) is the output power left at the termination. The relation between efficiency,
leakage rate, and length is generally dependent on the desired radiation pattern and therefore on the aperture
distribution: referring to a uniform-section LWA, where z is independent of z, one has = 1 exp(2z L).
It should also be noted that, for narrow-beam applications, very high efficiency should require an extreme
prolongation of the line source; in practice, in LWAs it is typical to radiate around 90% or at most 95% of the
input power, and the remaining power at the termination is absorbed by a matched load to avoid a back lobe
of radiation due to the reflected wave. (Other considerations related to the efficiency and the illumination will
be discussed in a subsequent subsection.)
Once the efficiency is chosen, a fixed relation therefore exists between the length in wavelengths (L/0 )
and the normalized leakage constant z /k0 . For a uniform-section LWA, an inverse proportionality between L
and z is found:
10
LEAKY-WAVE ANTENNAS
where c is related to the desired efficiency (e.g., for 90% efficiency c = 0.185). For a nonuniform section, since
z depends on z, the relation between efficiency, length, and leakage rate depends on the chosen illumination
and is more complicated.
In order to achieve narrow beams in the elevation angle, the effective longitudinal aperture has to be sufficiently wide (usually several wavelengths), and this implies a rather low leakage rate. The half-power (3 dB)
beam width is directly related to the antenna length given in free-space wavelengths (L/0 ) through an
approximate relationship, which takes into account also the contribution of the scan angle (3):
where the proportionality factor a is dependent on the aperture distribution: it has the smallest value for a
constant aperture distribution (a 0.88) and increases for tapered distributions (typically being more than
unity) (3). From the previous expression, it is seen that, since cos M kt /k0 , the beam width is also expressible
as 2/(kt L). This means that the beam width is, as a first approximation, practically constant when the
beam is scanned away from broadside by varying the frequency for air-filled LWAs (where kt is independent of
frequency), but it changes for dielectric-filled LWAs (where kt depends on frequency).
The effective aperture is reduced for a fixed antenna length as the beam approaches endfire (where the
previous expression becomes inaccurate), and tends in practice to increase. It can be seen that for an
ideal semiinfinite uniform structure (i.e. an antenna aperture from z = 0 to z = L ), the beam width
is determined by the leakage rate only, since in this case it can be found that 2z /kt . Moreover, in this
situation the radiation pattern depends only on z and z and does not present sidelobes:
For finite antenna lengths, sidelobes are produced and the expression for R() is more involved. In general
the specifications on the sidelobe level are related to the choice of the aperture distribution, whose Fourier
transform allows the derivation of the radiation pattern (see subsequent section on this subject).
Uniform and periodic leaky-wave antennas: scanning properties. It is seen that the beams for
LWAs based on partially open air-filled metallic waveguides scan, in theory, an angular region from around
broadside (z /k0 0) towards one endfire (z /k0 1).
In practice, around broadside the structure works near the cutoff region of the closed waveguide, where
reactive effects due to mismatching are increasingly important. The leaky-wave values for z /k0 cannot then
be extremely low, and at the same time z /k0 tends to increase too much, adversely affecting the possibility of
focusing radiation at broadside.
As regards the behavior at endfire it is seen that, since z /k0 tends to unity asymptotically as the frequency
increases, in the unimodal range (where these structures are usually employed) the beam cannot closely
approximate endfire radiation in an air-filled LWA. A way of improving the angular scanning is to fill these
structures with dielectric materials. Thus, since in this case the normalized phase constant approaches the
1
square root of the relative permittivity as the frequency is increased (z /k0 r 2 ), the value z /k0 = 1 can
actually be approached in a much more restricted frequency range. It should however be noted that for such
dielectric-filled structures the beam width may change strongly as a function of frequency and therefore as the
pointing angle varies [see comments on Eq. (7)].
Moreover, it should be noted that in many leaky structures (such as dielectric and printed ones), as the
frequency is increased, the leaky-mode solution changes into a guided-mode solution through a complicated
transition region (2,22,23,24). In this frequency range, also called the spectral gap, the contribution of the
LEAKY-WAVE ANTENNAS
11
leaky wave to the field tends progressively to decrease, and generally the structure does not work well as a
LWA. The representation of the electromagnetic field in this situation becomes more involved (the leaky-wave
contribution is in general no longer strongly representative of the aperture field, and the space wave assumes
increasing importance), and additional analyses are required, whose discussion is beyond the limits of this
work.
As mentioned, while the uniform LWAs usually radiate only in the forward quadrant, within the limits
specified above, the LWAs derived from periodically modulated slow-wave guides can radiate all the way to
backward endfire in the lower frequency range.
The design principles for periodic LWAs are for the most part similar to those for uniform LWAs (1,2,3,4).
The main difference lies in the characterization of the fast wave, which is now associated with a Floquet spatial
harmonic of the periodic guide (1,2,3,4,10,24). One can see that if a uniform guide is considered whose operating
mode is slow (z /k0 > 1; e.g., a dielectric waveguide), and a longitudinally periodic discontinuity is properly
added (e.g., an array of metal strips or notches, etc., placed at suitable distances p), such periodicity furnishes
a field expressible as an infinite number of space harmonics (zn p = z0 p + 2n), where z0 is the phase
constant of the fundamental harmonic, which is slightly different from the original value z of the unperturbed
guide. With proper choices of the physical parameters, it is in general possible to make only one harmonic fast
(typically, the n = 1 one), so that it can radiate as a leaky wave (with an additional attenuation constant z ).
In this case, the phase constant of this fast harmonic can assume both positive and negative values (1
< z /k0 < 1), as a function of the parameters involved: in particular, as the frequency is increased, the beam
inclines from backward endfire towards the broadside. In general, however, periodic LWAs have difficulties in
working well in the broadside region, since in general for periodic structures there exists an open stopband
(24) where the attenuation constant rapidly increases, resulting in a widening beam width with mismatching
effects.
As the frequency is further increased after attaining broadside, the beam is then scanned also in the
forward quadrant. In periodic LWAs, depending on the choice of the design parameters, additional limitations
on the forward scanning behavior may exist when a second harmonic starts to radiate before the first harmonic
reaches its endfire, thus limiting the single-beam scanning range (2,3,24).
Leaky-wave arrays for pencil-beam radiation. If an increase of directivity in the cross plane is
desired, a simple improvement in LWAs based on long radiating slots can be achieved by a physical enlargement
of the transverse aperture (e.g., with a flared transition to enlarge the effective cross aperture). As said before,
a more efficient way to increase directivity in the cross plane is to use a number of radiators placed side by
side at suitable lateral distances, thus constituting a linear array: it is then possible to achieve radiation with
a focused pencil beam. In addition, if properly phased, these arrays of LWAs allow 2-D scanning of the beam:
in the elevation plane, as is typical for LWAs, the scanning is achievable by varying the frequency, while in the
cross plane the scanning is achievable with phase shifters that vary the phase differences among the single line
sources. As noted, in LWAs only a unidimensional set of phase shifters is therefore necessary, yielding structural
simplicity and economic advantage in comparison with the usual radiators requiring a two-dimensional set of
shifters for the scanning. Additional desirable features of such arrays are in general the absence of grating
lobes and blind spots and the achievement of good polarization properties.
For the analysis of such LW arrays, an efficient method is that based on the unit-cell approach (11,12). In
this way, it is possible to derive the behavior of the whole structure by referring to a single radiator, taking into
account the effects on it due to the presence of all the others. In the equivalent network this is achievable by
changing only the description of the radiation termination for a periodic array environment (infinite number of
linear elements); in particular, an active admittance can be calculated, which describes the external radiating
region as a function of the geometry and of the scan angle. More sophisticated techniques also allow accurate
analyses of arrays by taking into account the mutual couplings for a finite number of elements (11).
Procedures for pattern shaping. Most of the basic considerations concerning the beam direction
and the beam width have been described in connection with a general line source, whose characteristics are
12
LEAKY-WAVE ANTENNAS
supposed to be longitudinally constant (uniform section, so that z and z are independent of z). However,
among the basic requirements on the radiation pattern, in addition to the specification for the maximum of the
beam direction and for its half-power width, the sidelobe behavior also has great importance. More generally,
it is desired to examine the properties of the source in comparison with a desired radiation pattern. Since
LWAs can be viewed as aperture antennas with a current distribution having a certain illumination A(z), it is
possible to obtain the far field through a standard relationship:
The radiation pattern for E is expressed in terms of a Fourier transform of the line-source complex current
distribution on the aperture multiplied by the pattern of the element current G (e.g., a magnetic dipole).
It is easily seen that if the LWAs geometry is kept longitudinally constant, the amplitude distribution
has always an exponential decay of the type exp( z z). As is known, this behavior furnishes a quite poor
radiation pattern for the sidelobes, which are rather high (around 13 dB). It follows that, in conjunction
with the choice of a fixed illumination function A(z) giving a desired sidelobe behavior (cosine squared, cosine,
triangular, Taylor, etc.), the leakage rate has to be modulated along the main direction z of the line source; in
practice this is achievable by properly modifying the cross section of the structure along z, a procedure usually
known as tapering. Considering that, for a smoothly tapered antenna, the power radiated per unit length from
the antenna aperture is simply related to the aperture distribution [viz., dP(z)/dz = 2z (z)P(z) = c|A(z)|2 ], it
is possible to obtain a useful analytical expression for z (z) as a function of the amplitude A(z), the line-source
length L, and the efficiency (3,4,25):
From Eq. (10) it is also seen that if high efficiency (close to unity) is desired, z has to be increased to
extremely high values near the terminal section (as mentioned, in common practice efficiency does not exceed
90% to 95%).
In general, in the tapering procedure the longitudinal modification of the geometry should be made in an
appropriate way in order to affect only the leakage constant, keeping the phase constant the same (in pencilbeam applications, z should not depend on z, in order to have the correct pointing angle for each elementary
current contribution on the aperture).
The pattern-shaping procedure requires therefore the knowledge of the phase and leakage constants as
functions of the geometrical and physical parameters of the chosen structure, and this is achievable, as was
said, by finding the suitable complex eigensolution with numerical methods. Since the pattern shaping requires
a proper z distribution with z constant, the procedure is greatly simplified if it is possible to find geometrical
parameters through which the leakage and phase constants are varied as independently as possible. This
possibility is related to the topology of the waveguiding structure.
An example of tapering is sketched in Fig. 4 for a leaky structure, the so-called stepped LWA [Fig. 4(a)],
recently proposed for high-performance applications with well-controlled radiation patterns (26). On the basis
of a preliminary analysis of the effects of the geometry on z and z , it is possible to find the appropriate
longitudinal variation of parameters to obtain the desired z distribution with a fixed z . In Fig. 4(b) the
detailed behavior of the modulation in the height of the lateral steps is shown as a function of z for a desired
illumination (cosine type). A first specification of the imbalance of the steps, with their mean value kept
LEAKY-WAVE ANTENNAS
13
constant (dashed profile), modifies appropriately the longitudinal distribution of the leakage constant, leaving
the phase constant almost unchanged. A second change is advisable to compensate the phase nonlinearity,
which can have rather disturbing effects on the radiation patterns; in this topology it is possible to vary the
steps mean value slightly, with the previously fixed imbalance, to obtain the final valid profile (solid curves).
The corresponding radiation patterns are then illustrated in Fig. 4(c) and 4(d), for the single-part and the
double-part tapering procedures: Fig. 4(c) is a rather distorted pattern related to the nonoptimized tapering
(dashed profile), while Fig. 4(d) is a correct cosine-type pattern related to the optimized tapering (solid profile).
The tapering procedure can be performed numerically in an easy way from a TRT network representation of
the structure. The typical scanning behavior of these kinds of antennas is finally illustrated in Fig. 4(e) for a
beam pointed by varying the frequency.
It is seen that, for the control of the aperture distribution, the longitudinal variation of the cross section
occurs in a continuous and quite smooth fashion, so that it can be assumed that for each z it is possible to
calculate the local value of kz as if the structures were infinitely long. Therefore, the first step in the design
procedure requires the evaluation of the z and z values as a function of the physical parameters of a specific
structure, in particular the dimensions and the frequency. After this basic analysis, which gives the phase and
leakage constants for a specified structure, the general synthesis procedure starts from the desired radiation
characteristics. Thus, from the angle of the beam maximum one finds the value of z /k0 at a desired operation
frequency; this value has to be kept longitudinally constant. From the beam width, it is also fixed the relative
length of the antenna. From the efficiency and the sidelobe specifications, the aperture illumination A(z) is
chosen, and then it is possible to calculate the leakage distribution z (z) from Eq. (10). Once the longitudinal
phase and leakage constants are determined, with the help of the analysis technique the required LWA
geometry is completely achievable.
14
LEAKY-WAVE ANTENNAS
Fig. 4. Example of the tapering procedure for a LWA to achieve a required aperture distribution for pattern shaping: (a)
Reference structure of a stepped rectangular-guide LWA, relevant parameters, and equivalent network. (b) Longitudinal
modulation of the lateral steps (bl , br , versus z) for a cosine illumination function for a microwave application. The leakage
mainly depends on the imbalance b = (b1 br )/(b1 + br ) of the steps, while the phase mainly depends on the their mean
height bm = (bl + br )/2. The dashed curves of bl , br versus z are obtained with a single-part tapering procedure, that is,
only varying b with constant bm (thus, variations on the phase constant are introduced). The solid curves are due to a
double-part tapering procedure, where phase errors are compensated by suitably varying bm : this is the correct profile for
achieving a radiation pattern with a cosine distribution [see part (c)]. (c) Distorted normalized radiation pattern R (dB)
according to the dashed curves. (d) Correct radiation pattern according to the solid curves for cosine illumination of the
stepped LWA. (e) Typical scanning properties for the pointed beam as a function of the frequency (stepped LWA).
5(d)] (31). In this case, the leakage is derivable from the perturbed dominant TE11 or other higher modes. Again,
practical limits may be encountered on the control of the leakage amount and on finding suitable parameters
for effective tapering.
Inductive-grid and asymmetrical trough-guide LWAs. The inductive-grid LWA [Fig. 6(a)] consists of a
parallel-plate guide operating in its first higher mode (TE1 ), with a cover made of closely spaced transverse
LEAKY-WAVE ANTENNAS
15
Fig. 5. (a) Slitted-side-wall rectangular-guide LWA; (b) channel-guide LWA; (c) holey- and slotted-side-wall LWAs; (d)
slitted circular-guide LWA.
metal strips similar to an array (32). This geometry allows for sidelobe control by varying the strip width to
control the leakage, and the side-plate height to adjust the phase. Though such a structure can work well,
delicate manufacturing problems can occur that make the geometry not very practical.
Another possibility based on a metallic guide, with wider apertures, is given by the trough waveguide
[(Fig. 6(b)] (33), which basically consists of a bisected strip-line structure operating in the first higher mode,
which becomes leaky when a suitable geometrical asymmetry (as a metal insert) is introduced with respect
to the central horizontal plane. It has been found that the structure offers flexible performance, though the
leakage properties can again be difficult to control by tapering.
Top-wall slitted and stub-loaded rectangular-guide and groove-guide LWAs. Reduced leakage is
achievable by slitting the top wall of the rectangular guide, decreasing the current modification due to the
cut [Fig. 7(a)]. By shifting the cut with respect to the central vertical plane, it is possible to modify the leakage
rate; investigations have also been performed with tapered meander profiles for sidelobe control (34). It is
found that the needed alterations of the phase constant for a tapered profile strongly affect the accuracy of the
radiation pattern, which can be severely distorted. Furthermore, this geometry does not easily allow for compensation of phase errors with an additional independent geometrical parameter. Moreover, the introduction
16
LEAKY-WAVE ANTENNAS
of a simple cut in the top wall of the guide can compromise the polarization purity, due to the effects of
higher-order modes at the aperture.
A way of improving the polarization purity in the basic geometry of a top-wall slitted rectangular guide
is to use an aperture parallel-plate stub, able to reduce the contribution of the higher modes at the aperture,
which are below cutoff, while the dominant leaky wave travels unattenuated as a TEM-like mode at an angle
(35,36,37) [Fig. 7(b)]. Wide metal flanges, simulating an open half space on the upper aperture, can increase
the directivity of this type of LWA. Attention has to be paid to possible channel-guide effects for large values
of the stub height. The topology might allow a printed-type version of this type of LWA, by filling the guiding
structure with a dielectric medium (3,12). It is also worth noting that the geometry under investigation can be
viewed equivalently as an asymmetrical horizontally bisected groove guide. Instead of shifting the stub with
respect to the central vertical plane, another possibility with such geometry is to create an asymmetry through
a small metallic insert [Fig. 7(c)]. Theoretical and experimental investigations have demonstrated the utility
of these structures, though the limits described above are still present when tapering techniques are required
for pattern shaping.
Slitted asymmetrical ridge and stepped rectangular-guide LWAs. The dominant mode of ridge waveguides can be made leaky with a symmetrical stub-loaded aperture if some asymmetry in the height of the
lateral arms is produced [Fig. 8(a)]. The characteristics of this topology show that the average value of the
heights (mean height) of the lateral arms of the ridge acts mainly on the phase constant, while the difference
between the arms heights (imbalance) acts mainly on the leakage constant (38). A tapering procedure that
first adjusts the leakage constant to set the imbalance and then compensates the phase constant through slight
changes in the mean height should be very simple and effective for pattern shaping.
The recently proposed geometry of the stepped LWA (26,39,40) [Figs. 4(a) and 8(b)] maintains the basic
advantages of the ridge structures with additional general desirable features (increased geometrical flexibility,
LEAKY-WAVE ANTENNAS
17
Fig. 7. (a) Top-wall-slitted rectangular-guide LWA; (b) stub-loaded rectangular-guide LWA; (c) strip-loaded groove-guide
LWA.
compactness, low profiles for aerospace applications, etc.). The structure has been investigated theoretically as
concerns the possibility of suitably adjusting the lateral steps for good independent control of phase and leakage. Simple design procedures show the actual capability of synthesizing arbitrary illumination functions with
greatly reduced sidelobes. It has been found that a two-part tapering procedure (first varying the imbalance
longitudinally and then compensating with slight modifications of the mean height) is generally sufficient for
appropriate sidelobe control. (An example of modulation by variation of the lateral steps has been presented
in Fig. 4.) Practical aspects have also been examined, including the effects of ohmic losses (25) and of radomes
for environmental protection (41). Experimental analyses have been performed, too, on both nontapered and
tapered prototypes (cosine illumination, etc.), for both microwaves and millimeter waves. Measured characteristics show very good agreement with theory and excellent radiation performance (precise pattern shaping,
high efficiency and power handling, good polarization purity, etc.) (39,40). Many practical advantages make
this structure one of the most desirable candidates among LWAs based on metal guides.
18
LEAKY-WAVE ANTENNAS
Fig. 8. (a) Slitted asymmetrical ridge LWA; (b) stepped rectangular-guide LWA.
printed waveguides that can act as leaky-wave radiators (coplanar guides, slot and strip lines, etc.) (12,42),
we particularly mention structures derivable from lengths of microstrip. Many different configurations can
be employed with microstrips acting as traveling-wave radiators. A first class is based on modulating the
dominant mode of the structure with periodic loadings, such as resonant patches or slots [Fig. 9(a)], and also
by varying the line shape periodically with different meander contours [Fig. 9(b)] (43). Different solutions have
been tested, but the theory of this topic seems to deserve further study.
A different configuration involves the use of uniform structures acting on higher-order modes that can
become leaky for certain values of the parameters involved [Fig. 9(c)]. The analysis of the complex propagation
characteristics of the microstrip line shows in fact that, in addition to the dominant quasi-TEM mode, the
higher-order modes generally become leaky in suitable frequency ranges (12,42) [see Fig. 2(a)]. In particular, it
is seen that the first higher mode EH1 can be excited with a proper odd-type source (the midplane of symmetry
is a perfect electric conductor) and, as the frequency is raised, it starts to leak power. In general, for the planar
structures, leakage can occur in two forms: the surface-wave leakage (power that is carried away through the
TE and/or TM surface modes of the layered structure), and the space-wave leakage (power that is carried away
through the TEM mode of free space). It is found that, for suitable choices of the parameters with an appropriate
excitation, the EH1 mode can represent rather accurately the radiation of the microstrip in a certain frequency
range (12) [see, e.g., Fig. 2(a)]. The coupling between the feeding and the radiating line must be well evaluated,
and simplified equivalent networks can be convenient for this purpose (45).
Radiation performance of printed-circuit LWAs (as concerns power handling, polarization, efficiency,
pattern shaping, etc.) is often less versatile and satisfactory than that of LWAs derived from metal guides
(12). From a practical point of view, difficulties are found particularly in adjusting the phase and leakage
constants independently through the physical parameters. Recent investigations involve simple possibilities
of tapering, also with microstrip topologies, utilizing other higher-order modes (46). Uniform-type microstrip
LWAs have also been investigated in array configurations for 2-D pencil-beam scanning (12,46,47).
LEAKY-WAVE ANTENNAS
19
Fig. 9. (a) Periodically loaded microstrip LWAs. (b) Periodic meander microstrip LWAs. (c) Uniform higher-mode microstrip
LWA: space-wave radiation can be associated, e.g., with the strip current distribution of the EH1 mode, which is leaky in a
suitable frequency range [see Fig. 2(a)].
Nonradiative dielectric guide LWAs. The nonradiative dielectric (NRD) waveguide is a recently proposed
guiding structure for millimeter-wave applications (48) [Fig. 10(a)]. It is a hybrid metaldielectric guide having
the same shape as the H guide proposed for microwave applications: it consists of a dielectric rod inserted
between metal plates placed at a distance apart that is less than the free-space wavelength. In this way, any
discontinuity that preserves the central horizontal-plane symmetry gives only reactive contributions, reducing
interference and radiation effects in integrated circuits. A number of passive and active components have been
realized with this topology (49,50), and integrated antennas have also been proposed (51,52,53,54). Usually,
NRD LWAs should employ some asymmetry in the basic geometry in order to make the operating modes leaky.
A first possible choice (51) [Fig. 10(b)] is to shorten the plates so that the bound operating mode (LSM01 )
(23,48,49,50) presents a nonnegligible amplitude contribution at the equivalent aperture and can give rise to
an outgoing leaky wave in the fast-wave range. Another possible choice (12,52) [Fig. 10(c)] is to insert some
geometrical asymmetry with respect to the central plane (typically an air gap between dielectric and metal), so
that a field having a net electric component perpendicular to the plates can be excited, and power can leak out
in the form of a TEM-like mode traveling at an angle in the parallel-plate region towards the outside. Various
analyses and design procedures have been developed for these configurations (51,52,53,54) in conjunction with
measurements on prototypes, both in single and in array configurations (12,53,54).
Dielectric-rod, image, insular, inset-guide, etc. LWAs. As was said, in basic dielectric guides a periodic
loading is required in order to isolate a suitable fast-wave space harmonic from the intrinsically slow-wave
structure. The reference structures can be various types of dielectric waveguides, such as circular or rectangular
20
LEAKY-WAVE ANTENNAS
Fig. 10. (a) Nonradiative dielectric (NRD) waveguide; (b) shortened NRD LWA; (c) asymmetrical NRD LWA.
dielectric-rod [Fig. 11(a)] (55), image (56), insular, and inset guides (57), [Fig. 11(b)]. The most usual periodic
perturbation is represented by a grating of grooves (58) or metal strips (55,59), usually placed in or on the top
surface of the guide [Fig. 11(c)]; also lateral metal patches can be used in hybrid forms (dielectricmicrostrip)
[Fig. 11(d)] (60). When sidelobe control is required, the taper is realized on the periodic perturbation (e.g.,
with grooves or strips slightly changing their dimensions longitudinally). Various studies have been made to
characterize the theoretical performance of these radiators (61); also, practical aspects have been analyzed,
such as the proper feeding elements in order to avoid spurious radiation and the reduction of the beam width
in the cross plane with flared horns (57) [Fig. 11(e)]. All these topologies are good candidates, particularly for
high-frequency applications (millimeter and submillimeter waves), where the use of dielectric instead of metal
for guidance can reduce the loss effects.
Layered dielectric-guide LWAs. It has been observed that LWAs based on single dielectric layers [Fig.
12(a)], with a ground plane on one side, usually present quite high leakage values, with consequent weak
capability of focusing radiation. A significant improvement is achievable by using additional dielectric layers [Fig. 12(b)]; in particular, interesting analyses have been performed on substratesuperstrate layered
LEAKY-WAVE ANTENNAS
21
Fig. 11. (a) Circular and rectangular dielectric rods; (b) image, insular, and inset dielectric waveguides; (c) periodically
loaded dielectric LWAs; (d) hybrid dielectric microstrip (insular guide with patches) LWA; (e) dielectric LWA with a flared
horn to reduce the cross-plane beam width.
structures (62,63,64). By properly dimensioning the heights and the dielectric constants (usually the substrate
has lower permittivity than the superstrate), it is possible to excite with a simple element (dipole or slot) a
leaky wave giving a conical (due to the symmetries of the topology), highly directive beam (62,63). Recently, this
basic substratesuperstrate topology has been arranged to allow for a very focused pencil beam with a limited
22
LEAKY-WAVE ANTENNAS
Fig. 12. (a) Dielectric-slab waveguide. (b) Layered-dielectric LWA based on a substrate superstrate structure with a
dipole excitation. (c) High-gain LW arrays of widely spaced elements in a substrate superstrate structure: linear and
planar configurations (for the latter case, a top view is shown for a microstrip feeding network of widely spaced slot
elements on the ground plane of the substrate superstrate structure).
number of radiating elements in the form of a widely spaced array, exploiting an interaction between leaky
and Floquet modes [Fig. 12(c)] (65). The advantages of such configurations are in the possibility of achieving,
through very simple design procedures, good radiative performance (high directivity, absence of grating lobes,
etc.) with an array of a few spaced 1-D or 2-D elements, thus reducing the cost of the beamforming network
and suitably exploiting the greater interspace available (dual-polarization applications, etc.).
Nonreciprocal structures: gyrotropic LWAs. In general the use of complex media can give additional
potentialities to LWAs. In particular, the possibility of achieving a unique widely scannable directional beam
with a symmetrical source, in addition to other advantageous features, has recently been investigated for basic
nonreciprocal structures employing gyrotropic media (66,67). A reference structure is considered, represented
by a transversely magnetized grounded ferrite slab with a simple current line source placed symmetrically
LEAKY-WAVE ANTENNAS
23
Fig. 13. (a) Nonreciprocal gyrotropic waveguide based on a transversely magnetized grounded ferrite slab with a line
current source; (b) bisected parallel-plate ferrite-loaded LWA.
inside the layer and in the direction of the bias field [Fig. 13(a)]. In practice, realizations of this topology can
be obtained with a bisected parallel-plate structure [Fig. 13(b)]. It has been found that further improvements
of the radiation performance are achievable with multilayered gyrotropic configurations.
In these nonreciprocal structures the analysis is still possible by means of a suitable extension of the TRT
(68). By varying the frequency or the bias field, even in a uniform structure, it is possible to have only one
pointed beam that scans continuously from one endfire to the other, passing through broadside. It has been
demonstrated that this unusual behavior is linked to two branches of a dominant anomalous mode of the
nonreciprocal waveguide (66): one branch is a complex improper solution (a standard leaky mode, responsible
for forward radiation); the other one is a complex proper solution, responsible for backward radiation.
24
LEAKY-WAVE ANTENNAS
towards the radiating line. Proper locations of matched loads or terminations again have to be arranged in
order to reduce the amount of power not properly radiated.
Ohmic losses usually do not greatly affect the radiative performance (efficiency, etc.) of LWAs, since the
attenuation due to the leakage of radiated power is generally more important than the attenuation due to
dissipated power in the nonideal guiding structure (25). However, as the frequency increases, power loss can
be excessive, particularly for LWAs based on closed metal guides. Therefore, for millimeter-wave applications
the choice of open guides with dielectrics and limited use of metal is often advisable.
The general simplicity of LWA structures makes their manufacture usually easy to accomplish, though
various construction problems can arise, depending on the chosen topology and the frequency range. Simple
structures are particularly desirable at millimeter wavelengths, due to the reduced dimensions. On the other
hand, too simplified shapes often cannot allow good control of the radiation performance. In particular, a
delicate aspect concerns the usually small longitudinal modifications of the geometry related to tapering for
sidelobe control. In this case, accurate analyses on the fabrication imprecisions and tolerances have a particular
importance in order to not overwhelm the required geometrical variations for tapering, thus degrading the
improvements of the pattern shaping (40).
Measurement techniques. The radiation properties of LWAs can be tested experimentally through
different types of measurements, most of them applicable to aperture antennas (1,2,3,4,40,69,70). Some basic
parameters, such as efficiency, mismatching effects, etc., can be measured directly through the transmission
and/or reflection scattering parameters with a network analyzer. Radiation patterns and directivity properties
as a function of the observation angles ( and in the zenith and azimuth planes, respectively) can be measured
for various values of the frequency, with different techniques, at the aperture, in the radiating near field (Fresnel
region), and in the far field (Fraunhofer region) (40).
Measurements at the aperture are quite easy to perform, in particular for LWAs incorporating partially
open metal guides. The basic parameters to be determined in LWAs, from which a complete knowledge of the
radiative characteristics is achieved, are the phase and the leakage constants. A measurement of the field in
close proximity to the aperture can be achieved with a small pickup element (e.g., an electric dipole probe placed
parallel to the aperture electric field). The amplitude and phase of the signal received by the probe are thus
measurable through a network analyzer, with possible compensations related to the mutual coupling between
the current distribution at the aperture and the current probe element (69). With numerical processing of the
data, the distributions of the leakage and phase constants are easily calculated. Far-field patterns are also
immediately calculable from these aperture measurements (40).
Alternative measurements can be performed directly on the radiation pattern in standard fashions. If
for instance an anechoic chamber is available whose dimensions do not allow immediate evaluations in the
far-field region (r > 2D2 /), measurements for achieving the radiation pattern are also possible by calculating
the LWAs performance at intermediate distances typical of the Fresnel region (radiating near field). The
LWAs field can be measured through another reference antenna (e.g., a calibrated horn), and rotating supports
are useful in order to facilitate the measurements as a function of the observation angles and frequency.
From these measurements at intermediate distance, by exploiting the equivalence principle, it is possible to
easily reconstruct the far-field radiation pattern. When either the dimensions of the anechoic chamber are
sufficiently wide or a controlled free-space region is available for simulating a far-field (Fraunhofer) region,
measurements of the radiation patterns are immediately achievable, again with a reference antenna and
movable structures. Experiments on all the types of measurements, recently performed for a class of LWAs
(40), show good agreement among the various results.
LEAKY-WAVE ANTENNAS
25
BIBLIOGRAPHY
1. F. J. Zucker Surface- and leaky-wave antennas, in H. Jasik (ed.), Antenna Engineering Handbook, 1st ed., New York:
McGraw-Hill, 1961, Chap. 16.
2. R. E. Collin F. J. Zucker (eds.) Antenna Theory, New York: McGraw-Hill, 1969, Chap. 19 (by A. Hessel) and Chap. 20
(by T. Tamir).
3. A. A. Oliner Leaky-wave antennas, in R. C. Johnson (ed.), Antenna Engineering Handbook, 3rd ed., New York: McGrawHill, 1993, Chap. 10.
4. C. H. Walter Traveling Wave Antennas 1st ed., New York: McGraw-Hill, 1965; 2nd ed., Los Altos, CA: Peninsula
Publishing, 1990.
5. N. Marcuvitz On field representations in terms of leaky modes or eigenmodes, IRE Trans. Antennas Propag., AP-4:
192194, July 1956.
6. S. Barone A. Hessel Leaky wave contributions to the field of a line source above a dielectric slab, part II, Microwave
Res. Inst. Report R698-58, Polytechnic Institute of Brooklyn, 1958.
7. T. Tamir A. A. Oliner Guided complex waves, parts, I, II, Proc. Inst. Elec. Eng., 110: 310324, 325334, Feb. 1963.
8. T. Tamir Inhomogeneous wave types at planar interfaces: IIIleaky waves, Optik, 38: 269297, 1973.
9. L. B. Felsen N. Marcuvitz Radiation and Scattering of Waves, Englewood Cliffs, NJ: Prentice-Hall, 1973.
10. R. E. Collin Field Theory of Guided Waves, New York: IEEE Press, 2nd ed., 1991.
11. R. C. Hansen, Ed. Microwave Scanning Antennas 1st ed., New York: Academic Press, 1966; 2nd ed., Los Altos, CA:
Peninsula Publishing, 1985.
12. A. A. Oliner Scannable millimeter wave arrays, Final Report on RADC Contract No. F19628-84-K-0025, Polytechnic
University, New York, Sept. 1988.
13. T. Itoh Millimeter-wave leaky-wave antennas, Proc. Int. Workshop Millimeter Waves, Italy, Apr. 1996, pp. 5878.
14. I. Ohtera Diverging/focusing of electromagnetic waves by utilizing the curved leakywave structure: application to
broad-beam antenna for radiating within specified wide angle, IEEE Trans. Antennas Propag., AP-47: 14701475,
1999.
15. T. Itoh (ed.) Numerical Techniques for Microwave and Millimeter-Wave Passive Structures, New York: Wiley, 1989,
Chap. 3 (by J. R. Mosig), Chap. 5 (by T. Uwaro and T. Itoh), and Chap. 11 (by R. Sorrentino).
16. R. Sorrentino (ed.) Numerical Methods for Passive Microwave and Millimeter Wave Structures, New York: IEEE Press,
1989.
17. N. Marcuvitz Waveguide Handbook, New York: McGraw-Hill, 1951.
18. P. Lampariello A. A. Oliner New equivalent networks with simple closed-form expressions for open and slit-coupled
E-plane tee junctions, IEEE Trans. Microw. Theory Tech., MTT-41: 839847, 1993.
19. C. Di Nallo F. Frezza A. Galli Full-wave modal analysis of arbitrarily-shaped dielectric waveguides through an efficient
boundary-element-method formulation, IEEE, Trans. Microw. Theory Tech., MTT-43: 29822990, 1995.
20. C. Di Nallo F. Frezza A. Galli Sviluppo di una procedura BEM per la caratterizzazione di structure guidanti arbitrarie,
Proc. XII Riunione Nazionale di Elettromagnetismo, Cetraro Italy, Sept.-Oct. 1998, pp. 277280.
21. L. O. Goldstone A. A. Oliner Leaky-wave antennaspart I: rectangular waveguides, IRE Trans. Antennas Propag.,
AP-7: 307319, Oct. 1959.
22. P. Lampariello F. Frezza A. A. Oliner The transition region between bound-wave and leaky-wave ranges for a partially
dielectric-loaded open guiding structure, IEEE Trans. Microw. Theory Tech., MTT-38: 18311836, 1990.
23. C. Di Nallo et al. Properties of NRD-guide and H-guide higher-order modes: physical and nonphysical ranges, IEEE
Trans. Microw. Theory Tech., MTT-42: 24292434, 1994.
24. S. Majumder et al. The nature of the spectral gap for leaky waves on a periodic strip-grating structure, IEEE Trans.
Microw. Theory Tech., MTT-45: 22962307, 1997.
25. C. Di Nallo et al. Rigorous evaluation of ohmic-loss effects for accurate design of traveling-wave antennas, J. Electromagn. Wave Appl. 12: 3958, Jan. 1998.
26. C. Di Nallo et al. Stepped leaky-wave antennas for microwave and millimeter-wave applications, Ann. Telecommun.,
52: 202208, Mar. 1997.
27. H. Shigesawa et al. Coupling between different leaky-mode types in stub-loaded leaky waveguides, IEEE Trans. Microw.
Theory Tech., MTT-42: 15481560, 1994.
26
LEAKY-WAVE ANTENNAS
28. P. J. B. Clarricoats P. E. Green A. A. Oliner Slot-mode propagation in rectangular waveguide, Electron. Lett., 2: 307308,
Aug. 1966.
29. J. N. Hines J. R. Upson A wide aperture tapered-depth scanning antenna, Report 667-7, Ohio State Univ. Res. Found.,
Dec. 1957.
30. R. S. Elliott Serrated waveguidepart I: theory, IRE Trans. Antennas Propag., AP-5: 270275, July 1957.
31. V. H. Rumsey Traveling wave slot antennas, J. Appl. Phys., 24: 13581365, Nov. 1953.
32. R. C. Honey A flush-mounted leaky wave antenna with predictable patterns, IRE Trans. Antennas Propag., AP-7:
320329, Oct. 1959.
33. W. Rotman A. A. Oliner Asymmetrical trough waveguide antennas, IRE Trans. Antennas Propag., AP-7: 153162, Apr.
1959.
34. F. L. Whetten C. A. Balanis Meandering long slot leaky-wave waveguide antennas, IEEE Trans. Antennas Propag.,
AP-39: 15531560, 1991.
35. P. Lampariello et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part Itheory, IEEE
Trans. Antennas Propag., AP-46: 10321041, 1998.
36. F. Frezza et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part IIeffects of flanges
and finite stub length, IEEE Trans. Antennas Propag., AP-46: 10421046, 1998.
37. M. Tsuji et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part IIImeasurements,
IEEE Trans. Antennas Propag., AP-46: 10471055, 1998.
38. F. Frezza M. Guglielmi P. Lampariello Millimetre-wave leaky-wave antennas based on slitted asymmetric ridge waveguides, IEE Proc. Microw. Antennas Propag., 141: 175180, March 1994.
39. C. Di Nallo et al. Theoretical and experimental investigations on the stepped leaky-wave antennas, 1997 IEEE AP-S
Int. Symp. Dig., July 1997, pp. 14461449.
40. C. Di Nallo et al. Strutture lineari e/o planari del tipo pointed beam a scansione di frequenza, Final Report on Alenia
Contract No. 1324002617/D/00, La Sapienza University of Rome, Italy, Oct. 1997.
41. C. Di Nallo et al. Analysis of the propagation and leakage effects for various classes of traveling-wave sources in the
presence of covering dielectric layers, 1997 IEEE MTT-S Int. Microwave Symp. Dig., June 1997, pp. 605608.
42. H. Shigesawa M. Tsuji A. A. Oliner New improper real and complex solutions for printed-circuit transmission lines and
their influence on physical effects, Radio Sci., 31: 16391649, Nov.Dec. 1996.
43. J. R. James P. S. Hall Handbook of Microstrip Antennas, London: Peregrinus, 1989.
44. F. Mesa C. Di Nallo D. R. Jackson The theory of surface-wave and space-wave leaky-mode excitation on microstrip
lines, IEEE Trans. Microw. Theory Tech., MTT-47: 207215, 1999.
45. P. Burghignoli et al. An unconventional circuit model for an efficient description of impedance and radiation features
in printed-circuit leaky-wave structures, 1999 IEEE MTT-S Int. Microwave Symp. Dig., June 1999, pp. 179182.
46. P. Baccarelli et al. Improving performance in leaky-wave antennas based on uniform microstrip lines, Proc. 29th
European Microwave Conf., Oct. 1999, pp. 126129.
47. C. N. Hu C. K. C. Tzuang Microstrip leaky-mode antenna array, IEEE Trans. Antennas Propag., AP-45: 16981699,
1997.
48. T. Yoneyama Nonradiative dielectric waveguide, in Infrared and Millimeter-Waves, K. J. Button (ed.) Vol. 11, New York:
Academic Press, 1984, pp. 6198.
49. T. Yoneyama Millimeter wave integrated circuits using nonradiative dielectric waveguides, Proc. Yagi Symp. on Advanced Technol., 1990, pp. 5766.
50. A. Galli Componenti e circuiti a microonde basati su risonatori dielettrici in guida NRD, Doctorate Thesis, La Sapienza
University of Rome, Feb. 1994.
51. A. Sanchez A. A. Oliner A new leaky waveguide for millimeter waves using nonradiative dielectric (NRD) waveguide
part I: accurate theory, IEEE Trans. Microw. Theory Tech., MTT-35: 737747, 1987.
52. A. A. Oliner S. T. Peng K. M. Sheng Leakage from a gap in NRD guide, 1985 IEEE MTT-S Int. Microwave Symp. Dig.,
June 1985, pp. 619622.
53. Q. Han A. A. Oliner A. Sanchez A new leaky waveguide for millimeter waves using nonradiative dielectric (NRD)
waveguidepart II: comparison with experiments, IEEE Trans. Microw. Theory Tech., MTT-35: 748752, 1987.
54. J. A. G. Malherbe An array of coupled nonradiative dielectric waveguide radiators, IEEE Trans. Antennas Propag.,
AP-46: 11211125, 1998.
LEAKY-WAVE ANTENNAS
27
55. M. Ghomi et al. Radiation characteristics of uniform and nonuniform dielectric leaky-wave antennas, IEEE Trans.
Antennas Propag., AP-41: 11771186, 1998.
56. T. Itoh B. Adelseck Trapped image guide for millimeter-wave circuits, IEEE Trans. Microw. Theory Tech., MTT-28:
14331436, 1980.
57. T. N. Trinh R. Mittra R. J. Paleta Horn image-guide leaky-wave antenna, IEEE Trans. Microw. Theory Tech., MTT-29:
13101314, 1981.
58. F. Schwering S. T. Peng Design of dielectric grating antennas for millimeter-wave applications, IEEE Trans. Microw.
Theory Tech., MTT-31: 199209, 1983.
59. S. Kobayashi et al. Dielectric-rod leaky-wave antennas for millimeter-wave applications, IEEE Trans. Antennas Propag.,
AP-29: 822824, 1981.
60. A. Henderson A. E. England J. R. James New low-loss millimeter-wave hybrid microstrip antenna array, Proc. 11th
European Microwave Conf., Sept. 1981.
61. M. Guglielmi A. A. Oliner Multimode network description of a planar periodic metal-strip grating at a dielectric
interfacepart I: rigorous network formulations, IEEE Trans. Microw. Theory Tech., AP-37: 534541, 1989.
62. D. R. Jackson N. G. Alexopoulos Gain enhancement methods for printed circuit antennas, IEEE Trans. Antennas
Propag., AP-33: 976987, 1985.
63. D. R. Jackson A. A. Oliner A leaky-wave analysis of the high-gain printed antenna configuration, IEEE Trans. Antennas
Propag., AP-36: 905910, 1988.
64. H. Ostner J. Detlefsen D. R. Jackson Radiation from one-dimensional dielectric leaky-wave antennas, IEEE Trans.
Antennas Propag., AP-43: 331339, 1995.
65. L. Borselli et al. Arrays with widely-spaced high-gain planar elements, 1998 IEEE AP-S Int. Symp. Dig., July 1998, pp.
14461449.
66. P. Baccarelli et al. The role of complex waves of proper type in radiative effects of nonreciprocal structures, 1997 IEEE
MTT-S Int. Microwave Symp. Dig., June 1997, pp. 491494.
67. P. Baccarelli et al. Improvements of the radiation performance in leaky-wave antennas with gyrotropic media, 1999
IEEE AP-S Int. Symp. Dig., July 1999, pp. 982985.
68. C. Di Nallo et al. A convenient transmission-line formulation for wave propagation in typical ferrite structures, IEEE
Trans. Magn., MAG-32: 32283236, 1996.
69. D. T. Paris W. M. Leach E. B. Joy Basic theory of probe-compensated near-field measurements, IEEE Trans. Antennas
Propag., AP-26: 373379, 1978.
70. T. S. Laverghetta Handbook of Microwave Testing, Dedham, MA: Artech House, 1981.
ALESSANDRO GALLI
FABRIZIO FREZZA
PAOLO LAMPARIELLO
La Sapienza University of Rome
388
LINEAR ANTENNAS
nications. This radiating piece of straight wire without curvature is the linear antenna. A simple example of a linear antenna is a two-wire transmission line carrying equal currents
in opposite directions and hence no resultant radiation. A
two-wire transmission line may be bent to create an efficient
radiator such as a dipole. The linear antennas have been
treated in numerous references. Some of them are in Refs.
113. We will describe key features of linear antennas in
this chapter.
(1)
(2)
1
E H)
Re(E
2
(3)
1
2
E H ) da
Re(E
(4)
LINEAR ANTENNAS
Historically, using a piece of radiating straight wire as an
aerial, or antenna, was a natural choice for wireless commu-
U = r2 Prad
(5)
The radiation resistance (Rr) is defined as the positive resistance across which the real power radiated (Prad) can be
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
LINEAR ANTENNAS
389
(6)
D =
4U
Prad + Prad
(13a)
D =
4U
Prad + Prad
(13b)
U d =
S
=0
2
=0
U sin d d
(8)
Prad
4
(9)
U
4U
=
U0
Prad
(10)
4Umax
Umax
=
U
Prad
(11)
Many practical antennas work with dual polarizations in mutually perpendicular directions, and then the directivity is defined in that particular direction; the total maximum directivity is a sum of directivities in mutually perpendicular
directions and is expressed as
D0 = D + D
G=
(14)
(7)
An antenna is a passive device, but it can be designed to radiate more energy in a desired direction. The gain (G) of an
antenna is defined as
(12)
Radiation efficiency () =
Real power radiated by the test antenna (Prad )
Total real input at the antenna feed terminals (Pin )
(15)
(16)
390
LINEAR ANTENNAS
For the infinitesimal dipole (Fig. 1), the current on the infinitesimal dipole is given by
J e (x , y , z ) = zJ
J 0
(18)
where
l/2
l/2
x
(a)
A(x, y, z) = z
Er
0
I dl
exp( jkr) dz = z 0 0
exp( jkr)
4r
4r
for r = 0 (excluding the source)
(19)
0 I0 dl
exp( jkr) cos
4r
I dl
A = Az sin = 0 0
exp( jkr) sin
4r
A = 0
Ar = Az cos =
(20a)
(20b)
(20c)
(b)
Figure 1. (a) The infinitesimal dipole and (b) its coordinate system.
This figure geometrically shows how the field at any observation point
from an infinitesimal dipole, which is a building block, can be estimated.
H=
Ar
1
(rA)
4r r
(21)
A= 0
4
+dl/2
dl/2
exp( jkR)
dz
J (x , y , z )
R
Hr = H = 0
H = j
(17)
1
k0 I0 dl
1+
sin exp( jkr)
4r
jkr
(22)
1
H
j
(23)
0 I0 dl
1+
2r2
k I sin
E = j0 0 0
4r
E = 0
Er =
1
cos exp( jkr)
jkr
1
1
1+
exp( jkr)
jkr
(kr)2
(24a)
(24b)
(24c)
LINEAR ANTENNAS
391
and
exp( jkr)
cos
2k0 r3
exp( jkr)
E
sin
= j0 I0 dl
4k0 r3
exp( jkr)
sin
H
= I0 dl
4r2
E = Hr = H = 0
Er = j0 I0 dl
r2
|E (r, , )|2
20
U=
(30)
(25a)
(25b)
Radiation Resistance
(25c)
(25d)
Several observations are in order. Er and E have (1/r2) variation as distance and therefore decays very fast. These are induction components and die down rapidly with distance. The
electric field components Er and E are in time phase, but the
magnetic field component H is in time quadrature with
them. Therefore, there is no time-average power flow associated with them. Hence, the average power radiated will be
zero, and the Poynting vector is imaginary. This can easily be
verified by integrating the average power density over a
sphere in the near region.
The space surrounding the antenna can be divided into
three regions, namely, induction, near-field (Fresnel), and farfield regions. The induction region has 1/r3 space variation,
the near field has 1/r2 variation, and the far field has a 1/r
variation with distance r.
Rr = 80 2
dl 2
(31)
P(r, , )
Far Field
exp( jkr)
sin
4r
Er = E = Hr = H = 0
E = j0 k0 I0 dl
H = jk0 I0 dl sin
exp( jkr)
4r
E
= m
H
z
y
(26b)
l/2
(26c)
l/2
(26a)
dz
=
x
(a)
P(r, , )
(27)
dz
l/2
l/2
z
z cos
Directivity
The radiation intensity U is given by
U = r Wav
2
(28)
where
Wav =
1
E H)
Re(E
2
(29)
x
(b)
Figure 2. (a) Thin linear antenna and (b) its coordinate system. This
figure geometrically shows how the field at any observation point can
be formulated using the basic building block, namely the infinitesimal dipole.
392
LINEAR ANTENNAS
/4
/4
I0
zI
0 sin[k(l/2 z )],
0 z l/2
zI
0 sin[k(l/2 z )],
l/2 z 0
(32)
0.707I0
/4
I
/4
I x (x = 0, y = 0, z )
/4
/4
t=0
t = T/8
t = T/8
(a)
(b)
(c)
/4
0.707I0
/4
I0
/4
/4
t = 3T/8
t = T/2
(d)
(e)
exp( jkR)
sin dz
4R
dEr
= dE = dHr = dH = 0
dE0 = j0 k0 Iz (x , y , z )
exp( jkR)
sin dz
dH
= jk0 Iz (x , y , z )
4R
Iin
I0 = Iin
where
R=
l<<
x2 + y2 + (z z )2 =
(33a)
(33b)
(33c)
with
(a)
l = /2
r 2 = x2 + y2 + z 2 ,
z = r cos
(b)
Iin
I0
Iin
I0
/2<l<
(c)
<l<3 /2
(d)
Figure 3. Current distribution on dipoles of different lengths. Different physical lengths of the dipole support different current distributions with varying number of half sinusoids. This is because at the
two ends of the wire, the boundary condition that the current must
be zero has to be satisfied.
(34)
Rr =
LINEAR ANTENNAS
(36)
W1 , L( f 1 )
W , L( f )
2
2
[Imn ] =
WN , L( f N )
J s (r ) G ds
1. Expand f as
n f n
(39)
n=1
n L( f n ) = g
W1 , L( f 2 )
W2 , L( f 2 )
WN , L( f N )
...
...
...
W1 , L( f 1 )
W2 , L( f 2 )
WN , L( f N )
(43b)
(40)
(41)
(44a)
W1 , g
W , g
2
[gN ] = . . .
...
(44b)
WN , g
n=1
is given by
1
2
[N ] =
..
.
N
(38)
f, g =
(43a)
(37)
L( f ) = g
N
(42)
N
n
Wm , L( f n) =
Wm , g
n=1
f =
393
exp( jkR)
dz = j / [B1 cos(kz) + B2 sin(kz)] (45)
4R
where
Iz(z ) the current flowing through an elementary length
of the wire
R distance of the observation point from the elementary length
394
LINEAR ANTENNAS
able in Richmonds work (17) and also in Ref. 8 are reproduced in Table 1.
2. Subdomain Expansion Functions. Subdomain expansions are attractive, convenient, and less expensive in
terms of computer time. This stems from the fact that
the current is matched on part of the integration path,
whereas for the entire domain the integration is performed over the whole path and for all N terms of the
expansions and coefficients determined.
Iz (z )
2
2
+ k G(z, z ) dz = jEzi
z2
(at = a) (46)
exp( jkR)
with R = a2 + (z z )2
4R
Ezi = the incident field (this is the source field for
antennas and scattered field for scatterers)
(47)
G(z, z ) = G(R) =
The availability of high-speed computers, graphics, and software packages, along with the enormous development of microcomputers, has made electromagnetic numerical computation extremely viable. An attractive feature of numerical
methods is their ability to handle arbitrarily shaped and electrically large bodies and bodies with nonuniformity and anisotropy where exact solutions can only at best provide some
physical insight. For large problems, it is possible to get a
linear system with a minimum set of equations to achieve a
certain accuracy.
An account of the area of numerical computation of thin
wire problem is well-documented in the literature (1,3,613).
As described in Ref. 8, there are many important computational issues involved in thin wire problems. These are (a)
segmentation of the structure and the correct number of segments, (b) choice of right current expansion functions, (c) the
thin wire approximation (radius a ), (d) Roundoff errror
due to matrix factorization, (e) near-field numerical anomaly,
(f) treatment of the junctions of the segments, (g) wire-grid
modeling, and (h) computer time required. Also, the errors (7)
involved are of concern. There are two types of errors encountered: (a) the physical modeling error, because in the absence
of an exact solution for a variety of semicomplex and complex
stuctures, it is the natural departure of the assumed structural details from the actual structure, and (b) the numerical
modeling error, since all numerical methods are approximate
but sufficiently accurate for the application.
A=
0
4
1
4
S (s)k(r) ds
I (s)S
(48b)
q(s)k(r) ds
(48c)
Formulation
Before we discuss the formulation of the thin-wire integral
equation, comments on the expansion functions used in this
study are in order. There are broadly two types of expansion
functions:
(48a)
Fourier
MacLaurin
Chebyshev
Hermite
Legendre
1
2
3
4
5
3.476
0.170
0.085
0.055
0.040
3.374
4.037
3.128
4.101
1.871
1.7589
1.5581
0.0319
0.0112
0.0146
8.2929
14.3644
4.4135
0.3453
0.0073
2.2763
2.1005
0.0655
0.0421
0.0322
LINEAR ANTENNAS
p1 ......... pn1
pn ......... pn+1
1 ......... n1
pN1
n+1 ......... N1
n+1 ......... N1
(a)
1 ......... n1
n
(b)
1
2
2
=0
exp( jkR)
d
R
1 dl
j ds
(49)
(50)
The MININEC solves the integral equation using the following steps:
1. The wires are divided into small equal segments such
that the length of the segment is still large compared to
the radius of the wire (a , typically 1/100th of a
wavelength). The radius vectors m, n 0, 1, 2, . . . are
defined with reference to a global origin.
2. The unit vectors are defined as
sn+1/2 =
r n+1 r n
|rr n+1 r n |
(51)
sn+1 + sn
,
2
sn1/2 =
sn sn1
2
I(s) =
N
In Pn (s)
(56)
Using this current expansion in Eq. (48b) and after mathematical manipulations which are available in Ref. 1 and are
not detailed here, we get the linear system matrix equation
in N unknowns:
[Vm ] = [Zmn ][In ]
(57)
sn+1 sn
sn sn1
This matrix has elliptical integrals involved in it. These elliptical integrals can be evaluated numerically.
The above equations are valid for any radius other than
small, for which the expression for breaks down and Harrington (18) provided an approximate formula for . This is
given by
=
=
s
k
1
ln
j 0
2
s
4
exp( jkrm )
4rm
for m = n
for m = n
(59a)
(59b)
When the pulse functions of Eq. (52) are inserted in parentheses, we obtain
s s
sm sm1
m
m+1
i
E (sm )
Sm1/2 +
Sm+1/2
2
2
sm sm1
Sm+1 sm
Sm1/2 +
Sm+1/2
= jA(sm )
2
2
(55)
n=1
The continuity equation given below determines the relationship between the charge q(s) and the rate of change of current
with distance:
q(s) =
395
m,u,v =
k0 (sm s ) ds
(60)
rn + rn1
=
2
(54)
396
LINEAR ANTENNAS
(61)
0.1
0.1
2m
2m
Wire 3
Wire 1
where
Wire 2
y
0.04 m
=
Zmn + Zload
for m = n
Zmn
for m = n
to meet the following requirements: (a) the segmentation density, (b) thin-wire criteria, (c) small radius calculation, (d)
step changes in wire radius, (e) spacing of wires, (f) loop antenna, and (g) monopoles and antennas above ground.
Operation of CurrentsLU Decomposition
Zmn
(a)
Wire 2
Overlapping pulse
(b)
Overlapping pulse
Wire 3
Wire 2
Wire 1
Overlapping pulse
(c)
Tee Antenna. Figure 7 shows the geometry of the Tee antenna fed from the base by a coaxial line. The impedance calculations of the Tee antenna using different computer programs including CURLU in MININEC and have been
compared with measurements (25).
Near and Far Fields. The near- and far-field programs
(FIELDS) calculate near and far fields using the current distribution on the structure obtained by integral equation formulations. The current distribution can be computed using
three programs: CURLU, CURTE, and CURRO. The current
distribution can be computed using perfect and imperfect
grounds, although the real ground corrections are applied to
the far field only. The real ground correction is included in the
form of reflection coefficients for parallel and perpendicular
polarizations. For details, the reader is referred to Chap. 8 of
Ref. 7. The menu is given below. User input(UI) means the
user is expected to respond at that point.
LINEAR ANTENNAS
104
100
103
MININEC
2
NEC
Antenna:
10.67 m monopole
0.102 m wire radius
20-Segment model
Near fields:
Magnitude of Ez and Er at z = 1.5 m
Radiated power = 8.775 mW at f = 1 MHz
101
10
102
MININEC
NEC
103
104
8
12
16
Horizontal distance (m)
NEC
105
20
MENU
1 - COMPUTE NEAR FIELDS
2 - COMPUTE FAR FIELDS
3 - SELECT/CHANGE ENVIRONMENT
4 - SELECT/CHANGE CURRENTS FILE
5 - SELECT OUTPUT DEVICE
6 - RETURN TO SYSTEM SUPERVISOR
0 - EXIT TO DOS
SELECTION (16 OR 0)? User Input
NAME OF INPUT CURRENT FILE? User Input (UI)
ELECTRIC OR MAGNETIC NEAR FIELDS (E/M)? User
Input
X-COORDINATE
Y-COORDINATE
INITIAL VALUE? UI INITIAL VALUE?
INCREMENT?
UI INCREMENT?
NO. OF PTS?
UI NO. OF PTS?
Z-COORDINATE
UI INITIAL VALUE? UI
UI INCREMENT?
UI
UI NO. OF PTS?
UI
PRESENT POWER LEVEL IS : CURRENT VALUE
CHANGE POWER LEVEL (Y/N)? UI
NEW POWER LEVEL (WATTS)? UI
Once the parameters are specified, the near- and far-field
results are printed out in words. Figures 8 through 15 show
the near-field characteristics of the monopole.
200
300
400
Radial distance (m)
500
102
Near fields:
H component at z = 1.5 m
Radiated power = 8.775 mW at f = 1 MHz
103
104
MININEC
105
NEC
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
100
MININEC
Er
Figure 9. Monopole near fields: Electric fields Ez and Er versus radial distance.
Ez
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
Measured
101
397
106
100
300
400
200
Radial distance (m)
500
398
LINEAR ANTENNAS
101
Near fields:
3
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
Er
Ez
50
100
MININEC
150
400
450
500
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
101
102
103
NEC
104
Antenna:
z - Component at wire surface
Radiated power =
8.775 mW at
f = 1 MHz
0.5
Near fields:
H component
at z = 1.5 m
Radiated power =
8.775 mW at
f = 1 MHz
1.5
50
100
150
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
36
108
72
Vertical distance (m)
144
180
Figure 13. Monopole near fields: Vertical electric field versus vertical distance.
COMPUTER CODES
A Summary of Different Available Codes
The following subsections summarize the scope of the codes
currently used for linear antenna analysis.
The Electromagnetic Surface Patch Code. The electromagnetic surface patch (ESP) code Version IV (26), which uses
the Method of Moments (MOM), is written for the analysis of
radiation and scattering from three-dimensional geometries.
These geometries include the interconnection of thin wires
and perfectly conducting and thin dielectric polygonal plates.
This code works for open as well as closed surfaces since the
formulation is based on electric field integral equation. It uses
polygonal plates modeling, which can generate an object as
complex as an aircraft with only 12 polygonal plates. The ESP
code is capable of doing the following:
0.5
105
400
450
500
LINEAR ANTENNAS
399
101
101
NEC
102
MININEC
MININEC
NEC
103
101
100
Near fields:
x-Component of E field
Radiated power =
8.775 mW at
f = 1 MHz
102
103
Near fields:
At x = 0.3 m
Radiated power =
8.775 mW at
f = 1 MHz
104
105
106
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
104
105
107
108
36
72
108
Vertical distance (m)
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
144
180
In order to use this code, the user specifies the (x, y, z) coordinates of the corners of each plate by entering the (x, y, z)
coordinates of the corners of each plate and the maximum
segment size in wavelengths for calculating the MOM modes.
The code automatically takes care of the frequency independence of the models.
In summary, the ESP code can treat a variety of geometries: (a) thin wires with finite conductivity and lumped loads,
(b) perfectly conducting or thin dielectric polygonal plates (30
wireplate junctions) at least 0.1 from the edge of a plate,
(c) plateplate junctions, including several plates of different
sizes which intersect along a common edge, and (d) excitation
by either a voltage generator or an incident plane wave.
ESP can compute the key characteristics of engineering interest of an antenna, namely, current distributions, antenna
input impedance, radiation efficiency, mutual coupling, near
or far zone radiation patterns for all polarizations, and near
or far zone back, bistatic, and forward scattering patterns
(full scattering matrix). There are still bugs in the ESP code,
and new features are being added to the codes. Nevertheless,
it should be possible to customize the code for a specific application.
The Mini Numerical Electromagnetic Code (MININEC). The
MININEC (20,21) is a computer program prepared in BASIC
language using the method of moments for the analysis of
linear antennas. It uses a Galerkin procedure (19) to solve
for wire currents using an integral equation formulation that
36
72
108
Vertical distance (m)
144
180
dz
l/2
z A
y
l/2
d = 2a
Figure 16. The center-fed thick dipole. A thick dipole has a frequency characteristic over a much wider band than a thin antenna.
The diagram shows how such a dipole will receive power from the
feeder network.
400
LINEAR ANTENNAS
1000
l/d = 25
(Moment method)
l/d = 50
(Moment method)
l/d = 104
(Sinusoidal current)
Input resistance ()
800
600
400
200
0
0.0
0.2
0.4
0.6
0.8
1.0
Dipole length (wavelengths)
1.2
1.4
1.6
1.2
1.4
1.6
(a)
400
Input reactance ()
200
l/d = 25
(Moment
method)
l/d = 50
(Moment
method)
l/d = 104
(Sinusoidal
current)
200
400
600
0.0
0.2
0.4
0.6
0.8
1.0
Dipole length (wavelengths)
(b)
Figure 17. The input impedance of wire dipoles with dipole length in wavelengths. (a) Input
resistance in ohms. (b) Input reactance in ohms. Every antenna presents an input complex impedance at its feed terminals. It is very important to know how the real and imaginary parts of
the impedance vary with parameters so that the antenna can be matched to the feeder.
LINEAR ANTENNAS
401
Iz (x , y , z )
a {I sin[k(l/2 z )] + jI0
z 0
=
z=h
z=0
Figure 18. The sleeve dipole. Apart from making the wire thick, another way of increasing the bandwidth of a thin wire is to put it in a
sleeve, which is basically a dielectric coating, and terminate by a
ground plane on one side.
(a)
(b)
(c)
(d)
[cos(kz ) cos(kl/2)]}
for 0 z l/2
[cos(kz ) cos(kl/2)]}
(62)
for l/2 z 0
for 0 |z | /4
The shape of the current does not change for half wave dipole,
but it does for other lengths, such as a wavelength.
402
;;
LINEAR ANTENNAS
18
16
14
G (mmhos)
12
10
Dielectric coated
dipole
L =
2a =
2b =
2 =
6
4
8"
0.025"
0.146"
2.360
2
0
0.1
0.2
0.7
1
b
ln
r
a
b
Q = (r 1) ln
a
P=
Calculated
Measured
at O.S.U.
0.8
(a)
10
(63a)
(63b)
BIBLIOGRAPHY
B (mmhos)
4
2
0
2
4
6
0
0.1
0.2
0.7
0.8
(b)
Figure 20. Comparison of measured and calculated admittance for
a coated dipole of length L 8 in.; 2a 0.025 in.; 2b 0.146 in.;
2 2.30. The figures show the bandwidth of antennas: (a) conductance and (b) susceptance of an insulated dipole vs. length in wavelengths.
ASOKE K. BHATTACHARYYA
Lincoln University
403
LOOP ANTENNAS
597
LOOP ANTENNAS
The IEEE Standard Definitions of Terms for Antennas (see
Ref. 1) defines the loop antenna as an antenna whose configuration is that of a loop, further noting that if the current
in the loop, or in the multiple parallel turns of the loop, is
essentially uniform and the loop circumference is small compared with the wavelength, the radiation pattern approximates that of a magnetic dipole. That definition and the further note imply the two basic realms of loop antennas:
electrically small and electrically large structures.
There are more than 200 million loop antennas currently
used by subscribers of personal communications devices, primarily pagers [see Ref. (2)]. More than a million a month are
currently being manufactured. Furthermore, loops have appeared as transmitting arrays, such as the massive multielement loop array at shortwave station, call sign HCJB, in
Quito, Ecuador, and as fractional wavelength-size tunable
high-frequency transmitting antennas. The loop is indeed an
important and pervasive communications antenna.
The following analysis of loop antennas reveals that the
loop, when small compared with a wavelength, exhibits a radiation resistance proportional to the square of the enclosed
area. Extremely low values of radiation resistance are encountered for such loops, and extreme care must be taken to
effect efficient antenna designs. Furthermore, when the small
loop is implemented as a transmitting resonant circuit, surprisingly high voltages can exist across the resonating capacitor even for modest applied transmitter power levels. The
wave impedance in the immediate vicinity of the loop is low
but at further distances (up to 2 wavelengths) exceeds the
intrinsic free-space impedance before approaching that value.
A loop analysis is summarized, which applies to loops of
arbitrary circular diameter and of arbitrary wire thickness.
The analysis leads to some detail regarding the current density in the cross section of the wire. Loops of shapes other
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
598
LOOP ANTENNAS
than circular are less easily analyzed, and are best handled
by numerical methods such as moment method described in
Ref. 3.
Loops are the antennas of choice in pager receivers and
appear as both ferrite-loaded loops and as single-turn rectangular shaped structures within the radio housing. Body-worn
loops benefit from a field enhancement due to the resonant
behavior of the human body with respect to vertically polarized waves. In the high-frequency bands, the loop is used as
a series resonant circuit fed by a secondary loop. The structure can be tuned over a very large frequency band while
maintaining a relatively-constant-feed point impedance.
Large loop arrays of one-wavelength-perimeter square loops
have been successfully implemented as high-gain transmitting structures at high-power shortwave stations.
(1)
D = 0 E = F
(2)
(3)
and with the vector identity () 0, where represents an arbitrary scalar function of position, it follows that
jF
F
H =
(4)
(5)
(6)
2
2
2
+ 2 + 2
2
x
y
z
(7)
(8)
h
S
I
(a)
(b)
2 + k2 = 0
(9)
Using Eq. (4) and the Lorentz condition of Eq. (6) we can find
the electric field solely in terms of the vector potential F. The
utility of that definition becomes apparent when we consider
a magnetic current source aligned along a single vector direction, for example, M zMz for which the vector potential is
F zFz, where z is the unit vector aligned with the z axis,
and Eq. (8) becomes a scalar equation.
LOOP ANTENNAS
1
Fz (r)
r2
r2 r
r
+ k2 Fz (r) = 0
(10)
(11)
ejkr
r
(12)
There is a second solution in which the exponent of the phasor quantity is positive; however, we are interested here in
outward traveling waves so we discard that solution. In the
static case the phasor quantity is unity. The constant C1 is
related to the strength of the source current and is found by
integrating Eq. (8) over the volume including the source, giving
0
kIS
4
C1 =
(13)
and the solution for the vector potential is in the z unit vector
direction,
F =
ejkr
0
kIS
z
4
r
Hr =
kIS jkr 2
e k
2
j
1
+
(kr)2
(kr)3
cos( )
j
kIS jkr 2
1
1
e k +
H =
+
4
kr
(kr)2
(kr)3
E = 0
kIS jkr 2
e k
4
1
kr
j
(kr)2
The Wave Impedance of Loop Radiation. The wave impedance can be defined as the ratio of the total electric field divided by the total magnetic field. We can study the wave impedance of the loop fields by using Eqs. (15) to (17) for the
infinitesimal loop fields, along with their dual quantities for
the ideal electric dipole. Figure 2 shows the loop field wave
impedance as a function of distance kr from the loop along
the direction of maximum far-field radiation. The wave impedance for the elementary dipole is shown for comparison.
At distances near kr 1 the wave impedance of loop radiation
exceeds 0 376.73 , the intrinsic free-space impedance,
while that of the infinitesimal loop is below 376.73 . In this
region, the electric fields of the loop dominate.
The Radiation Regions of Loops. Inspection of Eqs. (15) to
(17) for the loop reveal a very complex field structure. There
are components of the fields that vary as the inverse third
power of distance r, inverse square of r, and the inverse of r.
In the near-field or induction region of the idealized infinitesimal loop, that is, for kr 1 (however, r kS for the loop and
r h for the dipole), the magnetic fields vary as the inverse
third power of distance.
The region in which kr is nearly unity is part of the radiating near field of the Fresnel zone. The inner boundary of that
(14)
which is an outward propagating spherical wave with increasing phase delay (increasingly negative phase) and with amplitude decreasing as the inverse of distance. We may now solve
for the magnetic fields of an infinitesimal current element by
inserting Eq. (14) into Eq. (4) with Eq. (6) and then for the
electric field by using Eq. (2). The fields, after sufficient manipulation, and for r kS (see Ref. 10), are
sin( )
sin( )
(15)
10,000
2 Fz (r) + k2 Fz (r) =
599
1,000
100
Small loop antenna
(16)
(17)
where 0 c0 376.730313 is the intrinsic free-space impedance, c is the velocity of propagation in free space (see Ref.
13 for definitions of constants), and I is the loop current.
10
0.1
10
kr
Figure 2. Small loop antenna and dipole antenna wave impedances
compared. [Source: Siwiak (2).]
600
LOOP ANTENNAS
Z
V
E b J a H b M a ) dV =
(E
Z
V
E a J b H a M b ) dV
(E
(18)
That is, the reaction on antenna (a) of sources (b) equals the
reaction on antenna (b) of sources (a). For two loops with loop
moments parallel to the z axis we want to find the angle for
which the coupling between the loops vanishes, that is, both
sides of Eq. (18) are zero. The reference geometry is shown in
Fig. 3. In the case of the loop, there are no electric sources in
Eq. (18), so Ja Jb 0, and both Ma and Mb are aligned with
z, the unit vector parallel to the z axis. Retaining only the
inductive field components and clearing common constants
in Eqs. (15) and (17) are placed into (18). We require that
(Hrr H)z 0. Since r z sin() and z cos(), we
are left with 2 cos2() sin2() 0, for which 54.736.
When oriented as shown in Fig. 3, two loops parallel to the
xy plane whose centers are displaced by an angle of 54.736
with respect to the z axis will not couple in their near fields.
To be sure, the angle determined above is exactly correct for
Detector
Oscillator
infinitesimally small loops; however, that angle will be nominally the same for larger loops. Hazeltine (14) used this principle, placing the axes of the inductors in a common plane
each at an angle of 54.7 with respect to the normal form the
radio chassis, to minimize the coupling between the inductors.
The same principle can be exploited in the design of a
metal detector, as depicted in Fig. 4. The loop a is driven with
an audio frequency oscillator. Loop b, in a parallel plane and
displaced so that nominally 54.7, is connected to a detector that might comprise an audio amplifier that feeds a set of
headphones. Any conductive object near loop a will disrupt
the balance of the system and result in an increased coupling
between the two loops, thus indicating the presence of a conducting object near a.
The Intermediate- and Far-Field Zones of Loops. The loopcoupling problem provides us with a way to investigate the
intermediate- and far-field coupling by applying Eq. (18) with
Eqs. (15) and (16) for various loop separations kr. In the farfield region only the H term of the magnetic field survives,
and by inspection of Eq. (16), the minimum coupling occurs
for 0 or 180. Figure 5 compares the coupling (normalized
to their peak values) for loops in parallel planes whose fields
are given by Eqs. (15) to (17). Figure 5 shows the coupling as
a function of angle for an intermediate region (kr 2) and
for the far-field case (kr 1000) in comparison with the induction zone case (kr 0.001). The patterns are fundamentally and significantly different. The coupling null at 54.7
is clearly evident for the induction zone case kr 0.001 and
for which the (1/kr)3 terms dominate. Equally evident is the
far-field coupling null for parallel loops on a common axis
when the 1/kr terms dominate. The intermediate-zone cou-
0
kr = 0.001
kr = 2
90
270
kr = 1,000
b
y
a
x
Figure 3. Two small loops in parallel planes and with 54.736
will not couple in their near fields. [Source: Siwiak (2).]
180
(deg)
Figure 5. Normalized induction zone, intermediate zone, and far
zone coupling between loops in parallel planes. [Source: Siwiak (2).]
LOOP ANTENNAS
D(, ) =
1
4
E H ) r|
|(E
Pd =
(19)
Qloop
1.5I 2 Rr
= H2 0 =
4r2
kS k 2
4 r
(20)
for radiated power I 2Rr, hence, we can solve for the radiation
resistance:
Rr =
the Q, the quality factor defined in (2), is inversely proportional to the third power of the loop radius, a result that is
consistent with the fundamental limit behavior for small antennas.
If we use Eq. (22) and ignore the dipole mode terms and
second-order terms in a/b, the unloaded Q of the loop antenna is
E H ) r | sin( ) d d
|(E
(k2 S)2
0 = 0 (kb)4
6
6
(21)
6
8b
ln
2
a
=
(kb)3
(23)
3.6
(kb)3
(24)
which has the proper limiting behavior for small loop radius.
The Q of the small loop given by Eq. (23) is indeed larger than
the minimum possible Qmin (kb)3 predicted by Siwiak (2)
for a structure of its size. It must be emphasized that the
actual Q of such an antenna will be smaller than given by Eq.
(24) due to unavoidable dissipative losses not represented in
Eqs. (22) to (24). We can approach the minimum Q but never
go smaller, except by introducing dissipative losses.
The Gap-Fed Loop
The analysis of arbitrarily thick wire loops follows the method
in Ref. 8, shown in simplified form in Ref. 9 and summarized
here. The toroid geometry of the loop is expressed in cylindrical coordinates , , and z with the toroid located symmetrically in the z 0 plane. The relevant geometry is shown in
Fig. 6.
Loop Surface Current Density. The current density on the
surface of the toroidal surface of the loop is given by
XX
J =
601
Bn, p e jn Fp
(25)
n= p=
(22)
F1 = F0
2b
2 ab
a
2
cos( )
2
2b
1 (a/2b)
(26)
(27)
2a
(28)
LOOP ANTENNAS
1
4
Z Z
J cos( ) dS
(29)
1000
Radiation resistance,
602
and
A =
1
4
Z Z
J sin( ) dS
(30)
= 10
= 15
800
600
400
200
0.5
1.5
2.5
kb
j0
4k
Z Z
S
1 J
G dS
(31)
given by
where the value of dS [b a sin()]a d. The Greens function G is expressed in terms of cylindrical waves to match the
rotational symmetry of the loop,
1
G=
2j
jm( )
m=
Jm (1
(2)
v)Hm
(2
v)e
j (zz )
d
Hz =
(32)
where
v=
k2 + 2
1 = a cos( )
2 = + a cos( )
and where Jm() and H(2)
m () are the Bessel and Hankel functions.
Matching the Boundary Conditions. Expression (32) is now
inserted into Eqs. (29) to (31) and the electric field is then
found from Eq. (28) and the boundary condition is enforced.
For constant on the wire
E e j p d =
V0 sin(p)
p
H =
A
z
(34)
H =
A
z
(35)
A
A
1 A
+
(36)
The loop current across a section of the wire is found by integrating the function J in Eq. (25) around the wire cross section. The loop radiation impedance is then the applied voltage V0 in the gap divided by the current in the gap. Figure 7
shows the loop feed radiation resistance, and Fig. 8 shows the
corresponding loop reactance, as a function of loop radius kr
for a thin wire, 15, and a thick wire, 10, where
2 ln(2b/a) is Storers parameter (6). The thin-wire loop has
very sharp resonant behavior compared with the thick-wire
loop, especially for a half-wavelength diameter (kb 0.5)
structure. The higher resonances are less pronounced for both
loops. Thick-wire loops exhibit an interesting behavior in that
over a diameter of about a half wavelength, the reactance is
1500
(33)
= 10
= 15
Reactance,
1000
500
0
500
1000
1000
0.5
1.5
kb
2.5
LOOP ANTENNAS
Table 1. Parameter Y for Various Loop Thicknesses and
b 0.01 Wavelengths
19.899
17.491
15.294
12.886
10.689
8.2809
a/
0.000003
0.00001
0.00003
0.0001
0.0003
0.001
0.0039
0.0090
0.020
0.048
0.098
0.179
603
essentially always capacitive and the total impedance remains well behaved.
LOOP APPLICATIONS
Small Gap-Fed Loops. The detailed analysis of the thick,
gap-fed wire loop, as shown in Refs. 8 and 9, reveals that the
current density around the circumference of the wire, angle
in Fig. 6, is not constant. An approximation to the current
density along the wire circumference for a small diameter
loop is
J =
I
[1 2 cos()(kb)2][1 + Y cos( )]
2a
(37)
0 kI
2
(38)
I
2b
2
a
Zdipole = 0 (kh)2 j
6
kh
(41)
(39)
Loop antennas appear in pager receivers as both ferriteloaded loops and as single-turn rectangular shaped structures
within the radio housing. When worn on a belt the loop benefits from coupling to the vertically resonant human body. In
the high-frequency bands, the loop has been implemented as
a series resonant circuit fed by a secondary loop. The structure can be tuned over a very large frequency band while
maintaining a relatively-constant-feed point impedance. Onewavelength-perimeter square loops have been successfully
implemented as high-gain transmitting structures.
2a
2h
(40)
>> 1
Figure 9. A ferrite-loaded loop antenna. [Source: Siwiak (2).]
604
LOOP ANTENNAS
3 ln
Qdipole =
2h
1
a
(kh)3
15
12
9
(43)
0
10 log(QA /Q) = 40 dB
(44)
30 to 50
85
160
280 to 300
460
800 to 960
Source: After Siwiak (2).
6
3
4.5
(kh)3
Comparing the Q for a small dipole given by Eq. (43) with the
Q of a small loop of Eq. (24) we see that the loop Q is small
even though the same ratio of antenna dimension to wire radius was used. We conclude that the small loop utilizes the
smallest sphere that encloses it more efficiently than does the
small dipole. Indeed, the thin dipole, here masquerading as
the analog of a long thin ferrite loaded loop, is essentially a
one-dimensional structure, while the small loop is essentially
a two-dimensional structure.
We can use Eqs. (41) and (42) for the elementary dipole to
examine the ferrite-loaded loop antenna since it resembles a
magnetic dipole. The minimum ideal Q of this antenna is
given by Eq. (42), 1.01 106. The corresponding bandwidth
of such an antenna having no dissipative losses would be 2
35 MHz f/Q 70 MHz/1.01 106 69 Hz. A practical ferrite
antenna at this frequency has an actual unloaded QA of
nearer to 100, as can be inferred from the performance of beltmounted radios shown in Table 2. Hence, an estimate of the
actual antenna efficiency is
Frequency Band
(MHz)
Vertical
Horizontal
(42)
dB
Paging Receiver,
at Belt,
Av. Gain
(dBi)
Field Strength
Sensitivity
(dB V/m)
32 to 37
26
19 to 23
16
12
9
12 to 17
13
10 to 14
10
12
18 to 28
3
10
100
1,000
Frequency (MHz)
10,000
(45)
(46)
(47)
LOOP ANTENNAS
/4
0.14 to 0.25
/4
XC QP
(48)
2 Irms XC
(49)
Vp =
605
by recognizing that
Vp =
Irms =
Reflector
element
Feed
Driven
element
Stub
tuner
P
Re{Zloop}
(50)
element tips severe enough to damage the antenna when operated at high power levels (10 kW) in a high-altitude (10,000
ft) shortwave broadcasting application in the 25 m band.
Moore sought an antenna design with no tips that would
support extremely high electric field strengths that caused
the destructive arcing. His solution was a one-wavelength-perimeter square loop, later with a loop director element as
shown in Fig. 11. The configuration exhibited no arcing tendencies, and a new shortwave antenna configuration was born.
As pictured in Fig. 11, the driven element is approximately
one-quarter wavelength on an edge. Actually, resonance occurs when the antenna perimeter is about 3% greater than a
wavelength. The reflector element perimeter is approximately
6% larger than a wavelength, and may be implemented with
a stub tuning arrangement. Typical element spacing is between 0.14 and 0.25 wavelengths. The directivity of a quad
loop is approximately 2 dB greater than that of Yagi antennas
with the same element spacing.
BIBLIOGRAPHY
(51)
606
LOSS-ANGLE MEASUREMENT
KAZIMIERZ SIWIAK
Motorola, Inc.
592
Probe
signal
Master
station
Remote
station
593
nal decreases and the trail decay rate increases, so the operating frequency should be as small as possible according to
this argument. Important practical considerations favor
higher frequencies, however. For example, the desired signal
may be reflected by the ionosphere at frequencies below approximately 30 MHz, and these lower-frequency signals are
subject to higher atmospheric noise and are unreliable. Also,
antenna size and cost increase as the operating frequency is
decreased. As a result of the opposing criteria, the operating
frequency of most meteor-burst systems lies between 40 MHz
and 60 MHz. Several key characteristics of meteor-burst systems and channels are presented in Table 1.
THE METEOR-BURST CHANNEL
Master
station
Remote
station
Master
station
Remote
station
4060 MHz
1005000 W
Up to 2000 km
80120 km
Vehicle tracking, environmental monitoring, two-way messaging
0.21.0 s
1000 bits/min
594
P(t) = P0 exp
t0
cult to model. One model that does reasonably well for overdense trails has the form (5)
P(t) = P0 exp
1 at + b
2
ln
c
,
at + b
t0
595
BIBLIOGRAPHY
1. J. Z. Schanker, Meteor Burst Communications, Norwood, MA: Artech House, 1990.
2. G. R. Sugar, Radio propagation by reflection from meteor trails,
Proc. IEEE, 1964, pp. 116136.
3. J. A. Weitzen, Meteor Scatter Communication: A New Understanding, in D. L. Schilling (ed.), Meteor Burst Communications:
Theory and Practice, New York: Wiley, 1993, pp. 958.
4. R. A. Desourdis, Jr., Modeling and Analysis of Meteor Burst
Communications, in D. L. Schilling (ed.), Meteor Burst Communications: Theory and Practice, New York: Wiley, 1993, pp. 59342.
5. C. O. Hines and P. A. Forsythe, The forward scattering of radio
waves from overdense meteor trails, Can. J. Phys., 35: 1033
1041, 1957.
6. A. Papoulis, Probability, Random Variables, and Stochastic Processes, 3rd ed. New York: McGraw-Hill, 1991.
7. J. A. Weitzen and W. T. Ralston, Meteor scatter: An overview,
IEEE Trans. Antennas Propag., AP-36: 18131819, 1988.
8. K. Brayer and S. Natarajan, An investigation of ARQ and hybrid
FEC-ARQ on an experimental high latitude meteor burst channel, IEEE Trans Commun., COM-37: 12391242, 1989.
596
METEOROLOGICAL RADAR
9. S. L. Miller and L. B. Milstein, Error correction coding for a meteor burst channel, IEEE Trans. Commun., COM-38: 1520
1529, 1990.
10. M. B. Pursley and S. D. Sandberg, Variable-rate coding for meteor-burst communications, IEEE Trans. Commun., COM-37:
11051112, 1989.
11. J. M. Jacobsmeyer, Adaptive trellis-coded modulation for bandlimited meteor-burst channels, IEEE J. Select. Areas Commun.,
10: 550561, 1992.
12. J. M. Jacobsmeyer, Adaptive data rate modem, U.S. Patent No.
5,541,955, 1996.
13. E. J. Morgan, Meteor burst communications: An update, Signal,
42: 5561, 1988.
14. M. B. Pursley and S. D. Sandberg, Incremental-redundancy
transmission for meteor-burst communications, IEEE Trans.
Commun., COM-39: 689702, 1991.
15. M. B. Pursley and S. D. Sandberg, Variable-rate hybrid ARQ for
meteor-burst communications, IEEE Trans. Commun., COM-40:
6073, 1992.
16. S. L. Miller and L. B. Milstein, A comparison of protocols for a
meteor-burst channel based on a time-varying channel model,
IEEE Trans. Commun., COM-37: 1830, 1989.
17. C. W. Baum and C. S. Wilkins, Erasure generation and interleaving for meteor-burst communications with fixed-rate and variable-rate coding, IEEE Trans. Commun., COM-45: 625628,
1997.
18. S. Davidovici and E. G. Kanterakis, Performance of Meteor Burst
Communication Using Variable Data Rates, in D. L. Schilling
(ed.), Meteor Burst Communications: Theory and Practice, New
York: Wiley, 1993, pp. 383410.
19. D. L. Schilling et al., The FAVR Meteor Burst Communication
Experiment, in D. L. Schilling (ed.), Meteor Burst Communications: Theory and Practice, New York: Wiley, 1993, pp. 367381.
20. D. K. Smith and T. G. Donich, Maximizing throughput under
changing channel conditions, Signal, 43 (10): 173178, 1989.
CARL W. BAUM
CLINT S. WILKINS
Clemson University
MICROSTRIP ANTENNAS
Microstrip circuitry consists of a metal strip or patch on a dielectric substrate backed by a metal ground plane.
Microstrip antennas are becoming increasingly popular owing to their advantages in size, cost, conformity to
the supporting structure, low profile, and ease of fabrication. By using simple etching techniques, it is possible
to fabricate a wide variety of microstrip circuits, including arrays of antennas, feeding networks, and active
devices such that preamps or distributed transmitters can be conveniently placed next to the antenna elements.
In addition, diode phase shifter circuits can also be etched on the substrate to form single-board phased arrays.
This article describes the microwave properties of patch antennas.
The dimensions of microstrip patch antennas are large in comparison with the width of conventional
microstrip lines. The ma purpose of a patch antenna is to confine microwave energy to a small finite region.
Such confinement may be achieved with the resonant behavior of a finite guided structure supporting standing
modes. Electromagnetic energy radiates from that part of the patch that is open to free space, similar to a
slot cut along the side of a waveguide. To continuously radiate microwave energy away, the patch needs to be
electrically connected to a so-called feeder line. When the characteristic impedance of the feeder line matches
the impedance of radiation waves, power is directly dumped into the confined space of the patch without
causing much reflection. This means that all of the input power to the patch is radiated away into free space. A
conventional transmission line radiates little power, because the fringing fields alternate in sign over a short
distance, resulting in cancellation of their radiated field. In microstrip circuitry, power also radiates from open
circuits and from discontinuities, such as corners. The radiated power is small, however, because the radiation
impedances there are usually much higher than the characteristic impedance of the microstrip transmission
line encompassing these discontinuities.
This article introduces two calculation methods capable of describing antenna performance quantitatively.
The first method is less rigorous, but it has the advantage of being more analytical and hence can be applied
with ease. Called the resonant cavity model, it assumes perfect metal boundaries at the metal patch and
ground plane and magnetic-wall boundary conditions at the periphery, or sides of the patch antenna. It is
well known, that there are fringing fields at the periphery of the patch antenna. The influence of such fields
may be included in this approximate analysis by extending the linear dimension of the patch antenna by
a small amount consistent with the depth of the fringing fields. Radiation and material imperfections are
then considered as perturbations to the lossless cavity. As such, the resonant frequency, far-field pattern,
input impedance, radiation linewidth, and efficiency are all calculable. The only shortcoming of this simple
cavity model is that it is not able to address the possibility of the propagation of surface waves. Surface wave
generation becomes significant in comparison with patch radiation for a thick substrate with high dielectric
constant.
The second calculation method concerns Greens function analysis. In general, a Greens function is defined
as the solution of a differential or integral equation resulting from a (delta-function) point source satisfying the
required (hogeneous) boundary conditions. Thus, under arbitrary source excitation, the solutn of the equation
can be obtained by a superposition ofreens functions, still satisfying the boundary conditions. In the presence
of a point (dipole) current source in the background of a stratified structure consisting of dielectric or magnetic
1
MICROSTRIP ANTENNAS
layers, the excited electromagnetic field is termed a dyadic Greens function from the vectorial nature of the
source field and the observer field. Although the construction of the dyadic Greens function is straightforward,
this article pays special attention to its physical interpretations as it appeared in the original work by 1.
Actually, Sommerfeld solved for the first time the Greens functions associated with a horizontal point
dipole and a vertical point dipole in the presence of a semiinfinite conductor half-space: the Earth (1). Radio
waves are radiated into free space either directly from the point dipole source or indirectly from the image
dipole source induced by the (imperfect) ground plane, the Earths surface. These are called spatial waves,
because they exhibit a 1/r spatial dependence, where r denotes the distance from the original source point (or
image point) to the observer point. However, Sommerfeld also showed a second kind of radiation field, which
is tied to the airearth interface exhibiting a 1/ dependence, where denotes the 2-D distance between the
source point projected on the Earths surface to the observer point located also on the Earths surface (1). They
are called surface waves, because they decay exponentially when departing away from the airearth interface.
This article explains how the spatial wave solutions and the surface wave solutions are constructed from a
(horizontal) point dipole source in the presence of a layered dielectric or magnetic structure.
By using the dyadic Greens function, the general field solution in a stratified structure with external
current excitation can then be formulated in terms of an integral equation, which is subsequently solved by applying the numerical Galerkin method. This article outlines procedures to calculate engineering parameters of
general microstrip antennas and gives calculation results for circular microstrip antennas, including radiation
frequencies, bandwidths, far-field patterns, input impedances, effects of feeder-line position, and interference
between antennas. Recent developments on printed-circuit antennas are briefly reviewed, including the important broadband technique employing stacked parasitic elements to achieve high gain and low cross-polarization
levels. More sophisticated treatments on microstrip antennas may be found in Ref. 2.
Cavity Model
A microstrip patch antenna is a narrow-band device, and typically, the bandwidth covers approximately 5
percent of the radiation frequency. When the bandwidth, f , can be related to the Q of a resonator
it implies that power dissipation is not significant and the circuit of a microstrip patch antenna can be
approximated as a low-loss cavity resonator. This suggests that the performance of a microstrip patch antenna
can be analyzed using a perturbation method. That is, the zeroth order solution of the antenna is described
in terms of the eigen modes, or normal modes, of a lossless cavity. Losses are then added to the analysis as
first-order perturbations, including conductor loss, dielectric loss, and radiation loss. Magnetic loss can also be
included in the total loss if ferrites are used as the substrate material.
Analogous to radiation loss, surface wave loss may also be treated as a perturbation in the cavity model.
Radiation and surface wave losses may be represented as leaky waves from the antenna, the former directly
from air and the latter from the guided structure of the substrate. However, the analysis of surface-wave loss
may be difficult, because it involves Sommerfeld-type integrals where simple poles and branch cuts appear in
the complex k plane (1). Usually, surface waves are included only in a full-wave analysis in which Maxwell
equations are solved numerically in the microstrip geometry.
In contrast, the cavity model is easier to apply, allowing engineering parameters of a microstrip patch
antenna to be calculated analytically, including bandwidth, input impedance, radiation efficiency, and nearfield and far-field patterns. Furthermore, the physical meaning of normal modes is evident in the cavity
MICROSTRIP ANTENNAS
Fig. 1. Feeding the rectangular patch antenna by a microstrip line. The resonant cavity locates directly under the patch
bound by a magnetic wall, shown shaded in the figure.
model. For example, the left- or right-hand polarized radiation from a microstrip ferrite patch antenna can be
readily calculated by using the cavity model. The cavity model provides satisfactory quantitative answers if
surface wave losses are minimal. Surface wave losses are minimal for a thin substrate exhibiting low dielectric
constant, but not for a thick substrate with high dielectric constant. Also, for microstrip patches in proximity
to each other a full-wave analysis is required, since the cavity model is not sufficient to describe these complex
situations. The full-wave analysis involving the use of dyadic Greens functions is discussed later.
Resonant Frequency. Let a metal patch be deposited on top of a dielectric substrate backed by a ground
plane. The patch considered here is either of rectangular or circular geometry. Only dielectric substrates are
considered. Analysis of patch antennas with ferrite substrates may be found in Ref. 3. In the cavity model one
assumes a lossless substrate in which the cavity is bounded by an electric and a magnetic wall. The metal
surfaces of the microstrip patch and the ground plane may be approximated as perfect conductors or electric
walls, but the peripheral surface surrounding the antenna cavity directly under the patch is assumed to be a
magnetic wall. This is illustrated in Fig. 1, in which a microstrip rectangular patch antenna is represented by
a cavity with its periphery shown shaded as a magnetic wall. An electric or magnetic wall is defined such that
the tangential component of the electric or magnetic field vanishes at the wall boundary. Thus, a normal metal
boundary condition approaches that of an electric wall if the value of conductivity goes to infinity. A magnetic
wall imposes imaginary boundary conditions that insulate the inside of the cavity from the outside, allowing
no electromagnetic energy to propagate across it.
Under quasi-static assumptions a magnetic wall can be located at the periphery of a patch antenna,
provided that it encloses the total volume of the antenna plus its fringing fields. That is, the effective volume
of the cavity is slightly larger than the physical one, so that it accounts for both the electromagnetic energy
stored directly under the metal patch and its fringe. This is shown in Fig. 2, for the case of a circular microstrip
antenna, where the effective radius of the cavity is larger than that of the metal patch. Thus, the resonant
length at the side of a rectangular patch antenna, denoted as L = L + 2, is larger than the physical length,
MICROSTRIP ANTENNAS
Fig. 2. Feeding the circular patch antenna by a coax line. Due to the fringing-field effect the effective radius of the
resonator is larger than its physical value.
L, and the effective increment in length due to the fringing-field effect (4) is 2 with
where d is the thickness of the substrate, eff denotes the effective dielectric constant of the patch cavity (5)
given by
and r is the dielectric constant of the substrate material. The rationale behind Eq. (3) is that, depending on
the ratio of d/L, electric field lines can still propagate away from the cavity region, and so eff < r , and eff
r if d/L 1. The resonant frequency of the cavity resonator is therefore
where c denotes the speed of light in vacuum and n the order of the resonant mode (n = 1 for the fundamental
mode).
Based on the same fringing-field consideration, we expect that for a circular patch antenna the effective
resonant radius of the cavity, denoted as R = R + , is larger than the physical radius of the metal patch (6),
denoted as R, and
as shown in Fig. 2. A similar expression for eff as Eq. (3) is expected for a circular microstrip cavity. However,
such an expression is lacking in the present literature. We therefore use the value of r for eff , bearing in mind
MICROSTRIP ANTENNAS
that this approximation is true only when R d. The resonant frequency of the cavity is, therefore
where X mn is the nth zero of the derivative of the Bessel function J m (x) of order m.
We note that the above expressions for resonant frequencies for rectangular and circular microstrip patch
antennas [ Eqs. (4) and (6)] are derived based on the cavity model according to which standing modes are
excited within the cavity. At high frequencies the quasi-static assumption is no longer valid, and surface waves
of high orders are generated and propagating away from the cavity region guided by the dielectric substrate.
In this case, magnetic-wall boundary conditions are not appropriate at the periphery of the antenna cavity.
For these situations we must resort to a rigorous full-wave analysis to adequately address the leaky feature
of surface wavesfor example, by using the dyadic Greens functions to be discussed later. In the following we
use r for eff and L and R for L and R , respectively, with the difference understood.
Normal Modes and Feeder-Line Excitations. The normal modes excited in a microstrip patch antenna are usually transverse electric (TE) waves exhibiting no nodal points along the z axis, the direction
perpendicular to the substrate plane. Thus, the normal mode solutions show no z dependence, and the electromagnetic components can be uniquely solved from Ez , satisfying the following Helmholtz equation
where t denotes the transverse part of the del operator (with respect to the z axis), kmn is given by
mn is the angular frequency, and {m, n} refers to the index of the normal mode. The boundary condition
imposed on Ez is that the derivative of Ez along the normal direction of the boundary vanishes at the magnetic
wall. Once Ez is solved from Eq. (7), the corresponding H field is
For a circular cavity of radius R the normal modes are, [see Fig. 2]
MICROSTRIP ANTENNAS
and the modal wavenumber is determined by the zeroing condition of the derivative of Bessel functions of order
m
Here n denotes the unit vector pointing outward along the normal direction of the window, and h0 is a
constant specifying the excitation current. In Fig. 1 the excitation-current window is located on the shad area
directly under the microstrip feeder line with n being coincident with ey . The excitation field governed by
Eqs. (7) and (9) can be uniquely solved subject to the boundary condition of Eq. (14).
For the rectangular patch depicted in Fig. 1, let the microstrip feeder connection to the patch at the line
segment a x b, and y = 0. The excitation field therefore is
where
MICROSTRIP ANTENNAS
and
Note that at = mn , Am = , indicating that when the frequency of the external driving field equals one
of the normal mode frequencies, only that normal mode will be excited in the cavity responsible for radiation.
However, when losses are included, other modes with much smaller amplitudes can also be excited at resonance.
For the circular patch we consider the microstrip feeder line to join the metal patch at the arc
, and = R. The excitation field can be then written as follows
where
and and k are given in Eq. (19). Again, from Eqs. (25) and (26) it is seen that the coefficient of Cn goes to
infinity if normal mode frequencies are approached [Eq. (13)].
Let us consider the excitation of the patch antenna by a coax line directly under the patch. We may assume
the coax line to possess an inner filament of zero diameter ending in a point charge at the junction with the
patch (see Fig. 2). The excitation current density therefore is
MICROSTRIP ANTENNAS
In the presence of a driving current it appears as an inhomogeneous term at the right-hand side of Eq.
(7):
The boundary conditions are the same as those leading to normal mode solutionsthat is, electric walls
at metal boundaries and magnetic walls at the periphery of the patch cavity. Thus, Eq. (29) can be solved in
terms of normal mode solutions:
Here Eqs. (30) and (31) are for a rectangular patch and a circular patch, respectively, and the normal
modes, Ez mn z (x, y) and Emn z (, ), are given in Eqs. (10) and (12). Equations (30) and (31) imply that when k
approaches kmn , the normal mode {m, n} acquires an infinite amplitude. The corresponding magnetic field can
be derived using Eq. (9).
Thus, for the rectangular patch, we have
MICROSTRIP ANTENNAS
In Eqs. (32) to (34) and (35) to (37), kmn is given by Eqs. (11) and (13), respectively, and and k are defined
in Eq. (19). Once the excitation fields are known, Eqs. (16) to (18), (22) to (24), (32) to (37), losses of various
kinds, and hence the quality factor, Q, of the patch cavity can be calculated, as discussed later.
Input Impedance. Having solved the electromagnetic fields inside a lossless patch cavity, we can
relax the assumption of perfect electric and magnetic walls and allow electromagnetic energy to propagate
across these boundaries. This results in ohmic loss and radiation (surface wave) loss. Dielectric and magnetic
losses occur in the interior of the patch cavity, assuming a lossy medium exhibiting complex permittivity and
permeability, respectively. We consider first the radiation loss.
The KirchhoffHuygens principle, which is a vector analog of Greens theorem, states that the electromagnetic field inside a closed volume, V, can be derived by the volume charge and current distributions inside
V and the surface charges and currents distributed on the enclosing surface of V, denoted as S. The effective
10
MICROSTRIP ANTENNAS
electric surface current density, K e , magnetic surface current density, K m , and electric surface charge density,
where n denotes the unit outward vector normal to S, and is the permittivity. Thus, if we consider the outside
volume of the patch cavity as the volume V, we conclude that the effective magnetic current density on the
magnetic wall is
where n denotes the unit outward vector normal to the magnetic wall, with sign opposite that is Eq. (38), and
the factor of two accounts for the presence of the ground plane. K e and
do not appear on a magnetic wall.
The radiation field arising from the magnetic current density K m can be derived by using the same
formula describing the electric current density K e converted from the duality rule (7). The duality rule states
that the electromagnetic theory remains valid if all of the elic quantities are changed into the corresponding
magnetic quantities and the magnetic quantities are changed into the negative of the corresponding electric
quantities. Thus, the vector potential associated with K m can be written as (8)
where d denotes the thickness of the substrate, Cm the contour of the magnetic wall with counterclockwise
tangential unit vector t, e r the unit vector along r, a vector connecting the coordinate origin toward the
observation point, and
is the wavenumber in air. In Eq. (40) primed quantities originate at the source and assume |r r | d.
The fields in the far-zone are
where
MICROSTRIP ANTENNAS
11
where e and e denote unit vectors along the and directions at the observation point r. By using Eq.
(45) the far-field radiation associated with a circular patch antenna fed by a microstrip line, Eq. (22), is, for
example,
where it is understood that the ratio of sin m to m is 1 when m = 0. The far-field radiation pattern can be
calculated by using the following equation:
Conductor loss and dielectric loss can be derived with ease (7). The conductor loss is given by
where V c denotes the volume of the cavity and tan is the dielectric loss tangent.
12
MICROSTRIP ANTENNAS
The radiation or antenna efficiency is the ratio of radiated power to input power, or
Since, by definition, at resonance the stored magnetic energy, W m , equals the stored electric energy, W e ,
we have for the total stored electromagnetic energy
From the definition of Eq. (1), the total Q factor of the cavity is
upon which the voltage standing wave ratio (VSWR) bandwidth can be defined. We define Smax to be the
maximum value of VSWR that can be tolerated. We then have (9)
Typically, Smax = 2.
The input susceptance B, of the antenna can be calculated from the lossless patch cavity, which is a
zeroth-order quantity. However, the input conductance, G, must be calculated from the total loss of the cavity,
which is a first-order quantity. Thus
where I and V are the averaged input current and voltage experienced at the input feeder position. For a coax
feed with excitation current expressed by Eqs. (27) and (28), assuming a thin inner filament as shown in Fig.
2, I is known and V can be calculated from Ez evaluated at the input position multiplied by the thickness of
the substrate d. We therefore have
MICROSTRIP ANTENNAS
13
for rectangular and circular patch antennas, respectively. For a microstrip feed, the input voltage is obtained
according to Faradays law by averaging Ez over the feeder-line window joining the cavity multiplied by the
thickness of the substrate, d. The input current can be derived by applying Amperes law at the feeder-line
window, which is ho times the transverse length of the window, that is, ho (b a) for the rectangular antenna
and ho (2R) for the circular antenna. We therefore have
for rectangular and circular patch antennas, respectively. Here kmn in Eq. (60) is given by Eq. (13) and m in
Eq. (61) is given by Eq. (18).
From Eqs. (59), (60) and (62), we note that at the normal mode frequencies the input susceptance B =
0, and hence B1 = . These points are called antiresonance points as plotted in the Smith chart (see Fig. 3,
example for). The resonant points are defined to be purely resistive when the input resistance of the antenna
matches the feeder-line impedance, resulting in zero reflection. Therefore, B 1 = 0 at resonance. The resonance
point occurs slightly above the antiresonance frequency, which requires participation of normal modes of all
orders, although the normal mode responsible for antiresonance is excited with the greatest intensity. Thus,
by definition, at resonance the capacitive part of the stored energy equals the inductive part, rendering the
overall input reactance zero. In calculating the susceptance, we should also include the capacitance associated
with near-field excitation. However, its contribution has been accounted for as the increment in the effective
resonance length of the antenna, [See Eqs. (2) and (5)]. The input impedance is
We note that, in reality, at antiresonance, B is finite due to the presence of losses occurring at the antenna
patch cavity.
One severe drawback of the cavity model is that it is not able to address the surface-wave loss. 10 estimated
that surface-wave excitation is not important if d/0 < 0.09 for r = 2.3 and d/0 < 0.03 for r = 10, where
0 is the free-space wavelength. The criterion given by 11 is more quantitative: d/0 < 0.07 for r = 2.3, and
d/0 < 0.0023 for r = 10, if the antenna is to emit no more than 25 percent of the total radiated power as surface
waves. Recent work by Fonseca et al. showed that the size of the patch is also a parameter (12), as discussed in
more detail in the next section. When ferrite material is used as the substrate, magnetic loss can be estimated
14
MICROSTRIP ANTENNAS
Fig. 3. Integration contours for Sommerfeld integrals illustrating contribution from spatial wave and surface wave
excitations.
where denotes the imaginary part of the permeability. In concluding this section we note that the cavity model
depicts a semiempirical picture where the parameters have been adjusted to fit experiments, for example, the
effective dielectric constant eff , Eq. (3), and the increment in the resonant length of the patch, , Eqs. (2) and
(5). Nevertheless, the calculated radiation pattern and input impedance compare very well with measurements
(13).
MICROSTRIP ANTENNAS
15
context of a conventional Greens function method. The numerical technique applies only when the Galerkin
method is used to solve the resultant integral equations relating the unknown current variables to local electric
fields distributed across the metal patch boundaries.
The physical meaning of the Greens function is clear, and surface waves have been used with equal
importance as spatial waves in the Greens function formalism. Furthermore, the Greens function solution
usually requires 2-D calculations, in contrast to the generic 3-D computational methods. This is true when the
metal thickness is much smaller than the thickness of the substrate, as is usually the case. Material losses can
be readily included in the Greens function if complex permittivity and permeability values are used. It turns
out that efficient CAD tools can be constructed using the dyadic Greens function solutions, which calculate
engineering parameters of a microstrip antenna, including radiation frequency, far-field pattern, efficiency,
input impedance, and so on and analyze the cross-talk problem inherent in common microstrip circuitries.
To illustrate the physics in the application of a Greens function, we have decided not to regenerate
many mathematical formulas in this section. Instead, we concentrate on Sommerfelds approach to the Greens
function analysis (1), because it lends itself to a more physical understanding of the problem. We wish to
introduce the methodology leading to the formulation of the dyadic Greens function for a general stratified
structure, consisting of a finite number of dielectric and magnetic layers as constituents. We assume this layered
structure is infinite in both horizontal and vertical directions, although it is possible to include finite substrate
and radiation space by employing periodic boundary conditions, for example. Also, we assume the current
distribution is two-dimensional, resulting in 2-D analysis on the Galerkin elements. The finite conductivity of
the ground plane can be accounted for by invoking complex permittivity of the conductor layer (1). Results of
calculations are cited mainly from Ref. 15. Background material can be found in Ref. 16.
Point-Dipole Solutions. We start by introducing the vector potential, A, and scalar potential, V, in
electrodynamics subject to a Lorentz gauge (17), which has been implicitly used by 1
where J and are the current and charge densities, respectively, and normally = 0 (oscillating charges can
not easily be realized physically). We note that Eq. (68) does not specify A uniquely for a finite volume under
consideration (17). However, this gauge freedom is almost fixed for a system of infinite volume, because the only
source-free radiation for the entire space are the incoming waves from infinity, which can be readily checked
out and excluded from the solution of A by performing proper gauge transformation. From Eq. (67), the scalar
potential V is obtained from the divergence of A, and hence only the vector potential A needs to be solved.
We are now solving the vector potential, A(x, y, z), Eq. (68), induced by a point dipole current source, J 0 (x,
y, z), in the background of a dielectric or magnetic layered structure. A point dipole is also called a Hertzian
16
MICROSTRIP ANTENNAS
dipole, which can be approximated by a dipole antenna whose arms are much smaller than the wavelength of
radiation (1). Assuming the point dipole to be located at ( 0, 0, z0 ), the current density associated with the point
dipole is
where I 0 dl (= finite) denotes the strength of the dipole. Without loss of generality we assume I 0 is along the
x direction. From Eq. (68) we know that Ax is nonzero, corresponding to the radiation field from a horizontal
dipole. This is the only field that would be induced in empty space by the point dipole of Eq. (70). In the
presence of the layered structure, however, Az is also nonzero, because of the oblique reflection of Ax from
the layer interfaces, corresponding to the radiation field from a vertical dipole in the absence of the stratified
structure. Finally, Ay is identically zero as implied by the symmetry of the problem.
We require the tangential components of E and H to be continuous across the layer interfaces, which can
be directly written down from Eqs. (65) and (66). According to 1, these boundary conditions can be integrated
with respect to the transverse coordinates, x and y, and the constants of integration can be justified as zeros
by letting x and y go to infinity. For example, suppose one boundary condition requires the derivative of the
function f (x, y, z) with respect to x to be continuous across an interface. If f (x, y, z) vanishes as x goes to infinity,
either decreasing exponentially to zero as for a decaying wave or averaging to zero as for an oscillating wave,
we can integrate this boundary condition with respect to x and conclude that the function f (x, y, z) itself must
be continuous across the interface. As such, the boundary conditions imposed on the vector potential A = Ax e
x + Az e z as derived by Sommerfeld are the continuity conditions on the following four quantities (1)
where a, b, c, d are four unknowns to be determined by the boundary conditions and is given by
where and are the permittivity and permeability of the layer under consideration. For the uppermost layer
a = 0 = c and for the lower most layer b = 0 = d, as required by the boundary conditions at z = . For the
layer that contains the dipole source, we integrate Eq. (68) from z = z0 z = z0 + and note that the double
Fourier transform of the function (x) (y) is 1/2, and we derive the following discontinuity requirement on
MICROSTRIP ANTENNAS
x (kx ,
17
We therefore insert a fictitious interface at z = z0 assuming different values of a, b, c, d of for the two
subregions above and below this interface [Eqs. (72) and (73)], but with the same value of [Eq. (74)]. The
boundary conditions imposed on this fictitious interface are the same as before, except that the requirement
on the continuity of the quantity (Ax /z) is now being replaced by Eq. (75). If the z0 plane occurs at a single
layer interface of the stratified structure, no virtual interface needs to be created; we need only replace the
continuity requirement on (Ax /z) by Eq. (75).
Thus, we arrive at 4N unknowns with 4N boundary conditions for N interfaces, including the fictitious
one, if there is one. In the boundary conditions we replace [Eq. (71)] the operator /x by jkx , and /z by ,
whichever is applicable according to Eqs. (72) and (73). The 4N boundary conditions now become algebraic
ones and hence the 4N unknowns can be solved. The amplitudes of these unknowns are all proportional to the
dipole strength I0 d l, which can be conveniently chosen to be 1, as required by the Greens function. The vector
potentials Ax (x, y, z) and Az (x, y, z) can then be solved from x (kx , ky , z) and x (kx , ky , z) by applying the inverse
two-dimensional Fourier transforms. The electric and magnetic fields can finally be obtained by using Eqs. (65)
to (67).
Material imperfection results in losses of various kinds, upon which the permittivity and permeability
values become complex numbers. The dielectric loss is described by a loss tangent, tan, and the permittivity
takes the form of 0 r (1 jtan), where r denotes the dielectric constant. For a demagnetized magnetic
substrate, the permeability is 0 ( +j ) and and are the real and imaginary parts of the relative
permeability (3). For a metal conductor the permittivity contains both the displacement current and the
conduction current and hence the permittivity is modified as 0 j/, where denotes the conductivity. This is
the permittivity that was explicitly considered by 1, solving the electromagnetic fields generated by a Hertzian
dipole in response to the earths surface.
If a perfect metal is used as the ground plane, the electromagnetic field will not penetrate into it. Therefore,
the boundary condition on the metal surface is that the tangential components of the electric field vanish, and
it implies
Thus, we have 4N 2 unknowns and 4N 2 boundary conditions for a layered structure possessing a
perfect metal ground plane.
Actually, the interface boundary conditions [Eq. (71)] specify the flection and transmission of electromagnetic waves from one layer to another, and x is proportional to the magnetic field component and z is
proportional to the electric field component. Oblique-angle reflection and transmission can be readily written
down in terms of Snells law (18) and hence the coefficients of x and z , [Eqs. (72) and (73)] are determined
with relation to the preceding and succeeding layers. Thus, without the need of solving the boundary conditions
explicitly, [Eq. (71)], all of the a, b, c, d coefficients for the layers are correlated with each other and only four
unknowns remain, corresponding to those at the outermost layers (top and bottom), which can now be solved
by using the boundary conditions at the fictitious interface imposed by the point dipole [Eq. (75)], and so on.
For example, if the last layer is a perfect metal, the a, b, c, d coefficients of the layer adjacent to it satisfy the
following relationships
which are recognized as the total reflection condition. Equation (77) can be shown to be identical to Eq. (76).
18
MICROSTRIP ANTENNAS
For completeness, we list the transverse spectral-domain vector potential induced by a horizontal point
dipole located on top of a dielectric substrate backed by a perfect metal ground plane
Here the dipole is located at the plane z = 0, the dielectric substrate is of a thickness d and permittivity ,
and 0 and are given by Eq. (74) with subscript 0 referring to air. The zeros of Eqs. (80) and (81) correspond to
surface-wave TE and TM modes, respectively. While there exists at least one TM surface mode, it is not always
the case that TE surface modes will be excited. The threshold for TE mode excitation is
Sommerfeld Integrals. Within each layer the dipole field solutions are obtained by inverse Fourier
transforms of x (kx , ky , z) and z (kx , ky , z) over the two-dimensional kx ky plane, or, equivalently, k k
plane. Here (k , k ) denotes the polar coordinate and (kx , ky ) the Cartesian coordinate. In general
and
as can be checked from the boundary condition of Eq. (71). In other words, x is an even function of kx and z
is an odd function of kx , as implied by the symmetry of the problem. Therefore, after integration over k , we
MICROSTRIP ANTENNAS
19
obtain
Although, Eqs. (85) and (86) can be numerically integrated, as discussed later, it is instructive to discuss
a procedure outlined by 1. Actually, Eqs. (85) and (86) are termed Sommerfeld integrals. The most significant
discovery of Sommerfeld was that the seemingly real integration starting at the fixed point k = 0 can be
converted into a complex k integration over a path, W, that closes at infinity (See Fig. 4
where H 0 (2) are Hankel functions of the second kind of order 0 and 1, respectively. In Fig. 4, the contour W is
detoured slightly in the complex k plane in order not to run into branch cuts and poles such that the integrands
remain finite and single valued. Only two-layer substances are illustrated in Fig. 4, where the air is denoted by
the subscript 0 and a dielectric layer is denoted by the subscript 1. For example, the respective permittivities
for air and the dielectric layer are 0 and 1 (the permeability of the dielectric layer is assumed to be 0 , the
same as air). In Fig. 4, k0 and k1 are the branch points given by
and is the smallest zero assumed by the function DTM , see Eq. (81). It is understood that more branch points,
and hence more branch cuts, will appear in Fig. 4 if more layers are presented in the layered structure. Also,
more simple poles will appear if DTE and DTM of Eqs. (80) and (81) admit more zeros. In Fig. 4, the branch
point at k = 0 is associated with the Hankel functions H 0 (2) ; For Eq. (88), this branch point, and hence its
associated branch cut, is replaced by a simple pole due to the different singular behavior of H 1 (2) at the origin.
According to Sommerfield, the couour W shown in Fig. 4 can be analytically deformed into three or more
components surrounding the respective branch cuts and simple pole, denoted as Q0 , Q1 , and P. 1 showed that
contour integrals of Q0 and Q1 give rise to spatial wave radiation, and for a large (spherical) distance r, they
exhibit the following asymptotic dependence
Actually, Q0 is due to the free-space dipole radiation, and Q1 is the radiation wave diffracted by the
dielectric layer (or from the image dipole induced by that layer) (1). Assume the air-dielectric interface is
20
MICROSTRIP ANTENNAS
Fig. 4. Input impedance loci of the cited antenna. The solid line is from calculation, and small circles represent measurements made in Ref. 13.
located at z = 0. For z > 0, contribution form contour P has the following asymptotic forms
MICROSTRIP ANTENNAS
21
Fig. 7. Calculated radiation frequency as a function of substrate thickness (fed by microstrip line).
They are surface waves tied to the interface and decrease at a rate proportional to 1/ in the lateral
directions, in contrast to the spatial wave radiation exhibiting a 1/r dependence, [Eq. (90)].
In the presence of material losses, the branch points and simple poles associated with spatial and surface
waves, respectively, acquire imaginary components, which are then pushed off from the real axis toward the
lower half of the complex k plane. As such, numerical integration of Eqs. (65) and (66) can be properly carried
22
MICROSTRIP ANTENNAS
Fig. 8. Calculated radiation resistance as a function of substrate thickness (fed by microstrip line).
out. However, care needs to be taken to avoid large truncation errors. Integration of k along the real positive
axis can be distinguished in three regions
Region I 0 < k < k0
Region II k0 < k < kc
Region III kc <k <
Here, kc denotes a cutoff wavenumber to be discussed later. In Region I the integrands are well behaved.
However, at the resonant frequencies of a metal patch dictated by the cavity model, both the numerator and
denominator of the integrand vanish, although their ratio remains finite. We call these geometric resonant
points quasisingularities (15). Near these quasisingular points the numerator and denominator need to be
expanded in Taylor series upon which their common zeros cancel out.
All of the surface poles are contained in Region II, and, conventionally, the upper bound of Region II, kc ,
is defined to be 10 times the real part of the largest surface pole occurring at the integrand. In Region II the
integrand behaves wildly in the vicinity of a surface pole. When coming across a surface pole, the integrand
transits from positive infinity to negative infinity, resulting in sharp cancellation during numerical integration.
To circumvent this difficulty we expand the integrand in Laurent series in the vicinity of a surface pole kp = p,
and the quasisingular terms (with negative exponents) are then evaluated analytically (remember p is now a
complex number). Equivalently, the singular part of integration is obtained via residual calculations. After the
singular part is subtracted from the integrand, the integrand becomes regular and can then be numerically
integrated in Region II.
In Region III we are involved with integration of Bessel functions at infinity, which oscillate indefinitely
as infinity is approached without exhibiting a strict period. This renders the conventional extrapolation scheme
MICROSTRIP ANTENNAS
23
Fig. 9. Calculated radiation linewidth as a function of substrate thickness (fed by microstrip line).
inaccurate. To perform integration in this region we consider asymptotic expansion of Bessel functions
where denotes the gamma functions. As such, integrands are written in series containing terms of the
following form
which can be readily evaluated by exploiting sine and cosine integrals and their derivatives if p is a positive
integer, or error functions and their derivatives if p is a positive half-integer larger than 1 (15).
When gyromagnetic layers appear in the layered structure, instead of using the vector potential A, it
is more convenient to express the electric field E and magnetic field H directly in the formulation. This is
because the dispersion relation for a gyromagnetic medium together with the associated E and H fields are
already known (3), which can be readily used to compose the boundary conditions at the layer interface and the
metal patch position. We are still working in the transverse Fourier spectra domain but do not use the point
dipole solutions. Thus, as before, by starting from the outermost layer interfaces, we postulate the unknown
24
MICROSTRIP ANTENNAS
Fig. 10. Input impedance loci of the antenna for several coax feeder positions.
coefficients for Et and H t and translate the field components layer after layer until reaching the interface
containing the metal patch. For dielectric layers, the field solutions can be expressed in the form of Eqs. (72)
and (73) relating Et and H t on both sides of the layer. For a gyromagnetic layer, a similar expression is used,
except that the propagation constant in Eqs. (72) and (73) become nondegenerate; together with the four a,
b, c, d coefficients, these propagation constants need to be solved from a quartic dispersion relation (19). This
nondegeneracy for wave propagation with respect to different propagation direction and polarization comes
from the tensor behavior of the permeability, called Polder tensor (3). In principle, there is no fundamental
difference in handling dielectric and magnetic layers, except that more algebraic steps are involved in solving
the dispersion relation imposed by the Polder tensor. At the metal patch boundary the discontinuity in H t gives
rise to surface current densities th relates to Et through Ohms law and are expressed in the form of an integral
equation discussed later.
Numerical Solutions. The dyadic Greens function, G(r, r ), is defined as the electric field at location
r produced by a unit point dipole located at r . We can solve the current distribution over a microstrip metal
patch of negligible thickness deposited on a layered structure backed by a ground plane. Therefore, r and r are
both located at the metal patch, and essentially we have a two-dimensional problem with the third dimension,
the z-direction, being absorbed into the Greens function. Thus, we are required to solve the following integral
MICROSTRIP ANTENNAS
25
equation
where [ amn ] are the unknown coefficients to be solved expressed in terms of the current basis {J mn ()},
where I denotes the identity dyad. The inhomogeneous term of Eq. (96), [ bmn ], is associated with current
driving given by
26
MICROSTRIP ANTENNAS
where E mn is the electric field generated by J mn , J e is the driving current density, and integration is over the
whole volume, V. In Eq. (99) the reciprocity theorem has been used, which states that a response of a system
to a source is unchanged when the source and observer are interchanged (20). The unit drive-current density
may be specified as
Equations (100) and (101) are for coax and microstrip feeders, respectively, In Eqs. (100) and (101), S(x)
denotes the step function and
The radiation field associated with a point (Hertzian) dipole located at the air dielectric interface of a
microstrip structure is (16)
where
MICROSTRIP ANTENNAS
27
and
The radiation pattern for a given patch current distribution J(), Eq. (107), is therefore
where 0 = (0 /0 )1/2 and ko = 2/0 = (0 0 )1/2 are, respectively, the impedance and wavenumber in air.
A convenient current basis can be derived from the intrinsic functions associated with the metal patch
geometry (21). We define a set of current potential {mn ()} satisfying the two-dimensional Helmholtz equation
The boundary condition imposed on {J mn ()} is that the current is not allowed to flow across the metal
patch boundary, or, at the patch boundary
where n denotes the unit outward normal at the metal patch boundary. That is
Equation (112) determines the eigenvalues kmn in Eq. (109). The current potential can thus be determined,
and, for a rectangular patch located at 0 x L, 0 y W, we have
28
MICROSTRIP ANTENNAS
and mn denotes the root of the derivative of the Bessel function, J m (x)
The advantages of using current potentials are that not only the vector Galerkin equations are converted
into a scalar ones, but also most of the integrations can be carried out analytically, including those required
to perform inverse transverse Fourier transforms, rendering only one-fold Sommerfeld-type integrals to be
evaluated numerically. This is true even in the presence of multiple patches exhibiting close coupling. We note
that the current basis introduced in Eq. (110) corresponds to the normal-mode currents appearing in the cavity
model of the antenna resonator. Because the basis current is defined as the gradient of a scalar function [the
current potential, Eq. (110)], we call it an irrotational current (21).
In the following we present the results of some calculations for circular microstrip patch antennas (15).
For rectangular patch antennas, see Ref. 22. The first calculation applies to the published data of a circular
antenna characterized by the following parameters (13): R = 6.75 cm, d = 0.1588 cm, r = 2.62, and tan
= 0.001. The calculated resonance frequency of the fundamental mode is 0.7936 GHz (15), which compares
exactly with the measured value of 0.794 GHz (13). This is contrasted with the calculation for a cavity model
presented in the first section, which predicts a resonant frequency of 0.805 GHz (13). The calculated input
impedance of the antenna is shown in Ref. 15, which compares nicely with measurements shown as small
circles in Fig. 3 (13). Figures 5 and 6 show the calculated radiation pattern of the antenna in the = 0 and
= 90 planes, respectively. We note that only copolarized radiations are generated from the fundamental
mode excitation; the cross-polarized field cancels out for the two m = 1 and m = 1 modes at the fundamental
resonant frequency.
The second patch antenna geometry considered is a microstrip disk of radius R = 1 cm, which is fed by
either a coax line or a microstrip line (15). The substrate has of dielectric constant r = 2.2, loss tangent tan
= 0.001, whose thickness d that varies Figures 7 to 9, assuming = 0.2 rad, show the calculated resonant
MICROSTRIP ANTENNAS
29
frequency, input impedance, and radiation linewidth of the fundamental mode as a function of the substrate
thickness, respectively. In Fig. 7 the calculated resonant frequency of the antenna decreases monotonically
with the substrate thickness d, indicating that the effective dimension of the patch resonator increases with d,
as expected for a leaky cavity. Figure 8 shows that the input impedance of the antenna is relatively a constant,
unless d becomes very small, say, smaller than 0.05 cm. In Fig. 9, we see that the radiation bandwidth increases
with d, d hence the Q factor of the antenna decreases with d. This finding is consistent with Fig. 7 which exhibit
the leaky feature of the antenna patch cavity.
Figure 10 shows the calculated input impedance loci of the above circular antenna of substrate thickness
d = 0.1 cm fed by a coax line located at (0 , 0). In Fig. 10, the parameter re is defined as re = 0 /R, and re
= 1.0, 0.8, 0.6, and 0.416. For each feeder location, the two resonance frequencies shown in Fig. 10 are the
resonance frequencies of the probe-inductance in parallel with the detuned patch-resonator, forming a parallelresonant circuit, and the resonance frequency of the probe-inductance in series with the detuned patch-parallel
resonator, forming a series-resonant circuit. The field distribution of the patch at both frequencies shows very
little difference, and the patch is operated at the same resonance mode yet more or less detuned from its
resonance frequency, which occurs between these two frequencies. It can be seen from Fig. 10 that these
resonant frequencies appear only if 0.416 re 1. For re 0.416, no patch resonance can possibly be excited.
Figure 11 shows the calculated radiation frequency as a function of re , or 0 . It is seen in Fig. 11 that radiation
frequency remains roughly constant for 0.6 re 1, which increases rapidly when re is further reduced. Figure
12 shows the calculated input resistance as a function of re , or 0 . From Fig. 12, we see that the input impedance
decreases with re or 0 , and hence it is possible to design 50 input resistance of the antenna by feeding the
antenna at re = 0.442 (or re = 0.488 for 75 input impedance).
Finally, we consider the interaction between two identical circular microstrip patch antennas excited by
microstrip feeders of equal amplitude and phase. Let the antennas be deployed in parallel exhibiting the same
parameters considered before: R = 1 cm, d = 0.1 cm, r = 2.2, and the feeder lines having 50 resistance
(width 0.312 cm). The separation between the antennas is designated R 12 measured between their respective
30
MICROSTRIP ANTENNAS
Fig. 13. Calculated and measured input impedance loci of two coupled patch antennas.
centers. Figure 13 shows the calculated and measured input impedance for the coupled case, R 12 = 2 cm (the
patches are touching each other), and the uncoupled case, R 12 = 5 cm (the patches are separated by 3 cm at
their edges). Measurements were performed with respect to patch antennas fabricated using RT/Duroid 5880
material (Rogers Co., Chandler, AZ). The measured resonant frequencies were 5.514 and 5.561 GHz for the
coupled and uncoupled cases, respectively, which compare almost exactly with their calculated values of 5.5137
and 5.5642 GHz. The resonant frequencies of the patches and their input impedance have also been calculated
as a function of the patch separation, R 12 . These are shown in Figs. 14 and 15, respectively, with measurements
shown as solid squares. From Fig. 14, it is seen that the resonant frequency changes most rapidly when R12
is small, say, when 2 R 12 3 cm. Further increase in R12 does not change the resonance frequency much.
However, the input impedance does show long-range interference between two patch antennas. As shown in
Fig. 15 the input impedance is still increasing when the two antennas are separated by 5 cm, although the rate
of increase has slowed compared with its initial rate at R12 = 2 cm.
We conclude that the leaky feature of an antenna cavity can be well characterized by using a full-wave
analysis outlined previously. Among many numerical methods, the dyadic Greens function approach might
prove to be the simplest one to apply, not only because the analysis is two dimensional, but also because
the evaluation of the scalar Galerkin elements involves only onefold Sommerfeld-type integrals. Surface-wave
generation is significant for a thick dielectric substrate with large dielectric constant. The coupling between
MICROSTRIP ANTENNAS
31
Fig. 14. Calculated and measured resonant frequency of two coupled patch antennas as a function of their separation
distance.
microstrip elements is of long-range nature, although the radiation frequency of a patch is less influenced by
its neighboring antenna elements.
Recent Developments
Wideband Techniques. A microstrip patch antenna is inherently narrowband, stemming from the
cavity model that an electromagnetic resonator confining energy locally in space with little dissipation is
necessarily a high-Q, and hence narrowband, device. For a high-Q resonator the generation of surface waves
is insignificant. For a thin substrate, this condition amounts to d/0 < 0.07 for r = 2.3 and d/0 < 0.023 for
r = 10, as stated previously (10,11). For a microstrip antenna fabricated on a thin substrate the impedance
bandwidth is typically 1% to 3%. This is in contrast to the bandwidth of 16% of a half-wave dipole with a radius
to length ratio equal to 0.01, and 70% of a medium-length helix operating in the axial mode.
A number of techniques have been proposed to increase the bandwidth of a microstrip patch antenna.
These techniques are generally classified into three categories. The first category involves a straightforward
approach based on the use of a thick substrate whose dielectric constant is small. The second approach is
to design a matching network to enhance the bandwidth (23). The third method uses parasitically coupled
elements in a variety of ways to produce closely spaced multiple resonances of the antenna (24,25).
By using a thick substrate whose dielectric constant is considerably different from that of air, surface
waves will be generated and inevitably reduce the radiation efficiency and introduce interference between
array elements. To resolve this problem, air cavities or holes may be introduced with the substrate to effectively
reduce the dielectric constant of the patch. For example, 26 machined closely spaced holes in a Duroid substrate
underneath a microstrip patch to lower the effective dielectric constant of the antenna. Using a micromachining
technique, Zheng et al. measured a 12.8% impedance bandwidth on cavity-backed microstrip patch antennas
fabricated on silicon wafers (27).
32
MICROSTRIP ANTENNAS
Fig. 15. Calculated and measured input impedance at resonance of two coupled patch antennas as a function of their
separation distance.
The bandwidth of the antenna is primarily determined by the rate of transition that the imaginary
part of the impedance changes sign at resonance, as discussed previously. Thus, it is possible to introduce
a cancellation mechanism on the input inductance so as to smooth the impedance variation. For example,
the inductance associated with the long wire-lead of the coax probe will limit the bandwidth to <10% for a
thick substrate. By etching a small circular slot around the probe on the patch, capacitance is introduced that
cancels the probe inductance to produce a bandwidth of 16% (28). Recently, with the use of a U-shaped slot,
a substantial increase in bandwidth (32%) has been demonstrated (29). Alternatively, an L-shaped probewire
has been shown to result a bandwidth of 28 percent (30).
The third method to achieving broadband operation is to couple the microstrip patch antenna parasitically
with other dielectric resonators characterized by approximately the same resonant frequency of the patch. For
an isolated patch the resonant frequency is determined by its lateral dimension. In the presence of fringe fields
at the patch periphery, however, the boundary of the antenna is neither sharply nor rigidly defined, leading to
a slightly larger effective dimension of the patch, as described by Eqs. (2) and (5) for rectangular and circular
microstrip patches, respectively. Thus, in contrast to a metal cavity, patch antennas have soft boundaries, which
in turn give rise to finite bandwidth in radiation: the softer the boundary, the wider the bandwidth. For this
reason, a patch antenna fabricated on a thick substrate will show a wide bandwidth. Similarly, when coupled
together, many soft-boundary microstrip resonators or patches, the overall radiation bandwidth is consequently
enlarged.
The disadvantage of using a thick substrate is that, besides its cost, surface waves may be generated
in the substrate so as to reduce the antenna feeding efficiency. By using multiple electromagnetically coupled
patches this disadvantage can be overcome. Electromagnetically coupled patches can be deployed either side by
side (laterally coupled geometry) or layer by layer (vertically coupled geometry). For laterally coupled patches,
the antenna size will increase considerably, ultimately restricting its usage large-array applications (31).
MICROSTRIP ANTENNAS
33
The two-layer electromagnetically coupled patch antenna consisting of a driven (feeder) patch in the
bottom and a parasitic (radiating) patch on the top has been investigated by several authors. For circular
(32), equitriangular (33), and rectangular (34) patches, experimental results have shown an enhanced gain
and impedance bandwidth with low cross-polarization levels as compared with the conventional single-layer
microstrip antenna. Actually, like a Yogi antenna, the gain of a stacked antenna can be increased to above 20
dBi at any scan angle if the thickness of the substrate and multiple superstrate layers is chosen pperly (35).
Therefore, using stacked parasitic elements in microstrip arrays could improve the overall array performance,
offering a higher gain and broad bandwidth, meeting with the same array design criteria of a conventional
single-layer microstrip array but with fewer array elements.
Size Reduction. Microstrip antennas have a number of advantages over conventional antennas,
namely, small size, light weight, low production cost, and natural conformity. For many applications, however, for example, hand-held mobile communications systems, half-wave microstrip antennas etched on a
low-cost dielectric substrate are still too large to be accommodated on a portable phone. A well-known method
of reducing the size of a half-wave patch to a quarter wave is to introduce an electric-shorting wall at one of the
radiating edges. Hiraswa and Haneishi (36) have shown that the length of the patch can be made sufficiently
shorter than a quarter wave by replacing the shorting wall with a shorting pin at the corner of the patch.
Recently, Wong and Lin have shown that by replacing the shorting pin with a chip resistor of low resistance,
antenna size can be further reduced, with an additional increase in bandwidth (37).
The other approach to reduce antenna size is to use a meandered geometry of the patch antenna (36).
By meandering the patch, the effective electrical length is larger than the physical length. Consequently, the
resonant frequency of the meandered antenna can be much lower than that of a conventional design with the
same physical length (38).
Finally, we note that not only is it possible to achieve high gain for printed circuit antennas, it is also
feasible to shape the radiation pattern in a prescribed manner. Some very interesting phenomena, such as
radiation into the horizon, radiation pattern monodirectionality, and azimuthal-dependent radiation have been
found possible (39). Furthermore, laterally coupled patches can serve as adaptive array antennas, because the
phase of radiation from a parasitic element can be adjusted via a varactor diode inserted at some feeding
position (40). The important feature of an adaptive array antenna is that it can provide beam steering as
required by many communications and traffic control systems (41).
BIBLIOGRAPHY
1. A. Sommerfeld, Partial Differential Equations, New York: Academic Press, 1962.
2. J. R. James P. S. Hall (eds.), Handbook of Microstrip Antennas, Vols. 1 and 2, London: Peter Peregrinus, 1989.
3. H. How, T.-M. Fang, C. Vittoria, Intrinsic modes of radiation in ferrite patch antennas, IEEE Trans. Microw. Theory
Tech., MTT-42: 988, 1994.
4. E. O. Hammerstad, Equations for microstrip circuit design, Proc. 5th Eur. Microstrip Conf., Hamburg, September, 1975,
pp. 268272.
5. M. V. Schneider, Microstrip lines for microwave integrated circuits, Bell System Tech. J., 48: 14211444, 1969.
34
MICROSTRIP ANTENNAS
6. L. C. Shen, Resonant frequency of circular disc, printed circuit antenna, IEEE Trans. Antennas Propagat., AP-25: 596,
1977.
7. J. A. Stratton, Electromagnetic Theory, New York: McGraw-Hill, 1941.
8. R. S. Elliott, Antenna Theory and Design, Englewood Cliffs, NJ: Prentice-Hall, 1981.
9. A. G. Derneryd, Microsip array antenna, Proc. 6th Eur. Microw. Conf., 1976, pp. 339343.
10. J. R. James A. Henderson, High-frequency behavior of microstrip open-circuit terminations, IEE J. Microw. Opt. Acoust.
3: 205218, 1979.
11. C. Wood, Analysis of microstrip circular patch antennas, IEE Proc., 128 (H): 6976, 1981.
12. S. B. A. Fonseca A. J. Giarola, microstrip disk antennas, part I: Efficiency of space wave launching, IEEE Trans.
Antennas Propagat., AP-32: 561567, 1984.
13. Y. T. Lo, D. Solomon, W. F. Richards, Theory and experiment on microstrip antennas, IEEE Trans. Antennas Propagat.,
AP-27: 137145, 1979.
14. R. C. Booton, Jr., Computational Method for Electromagnetics and Microwaves, New York: Wiley, 1992.
15. H. How C. Vittoria, Directions for the next generation of MMIC devices and systems, in N. K. Das and H. L. Bertoni,
(eds.), Computer Aided Design Tools for Microstrip Circuitries, New York: Plenum, 1997, pp. 399406.
16. J. R. Mosig, R. C. Hall, F. E. Gardiol, Numerical analysis of microstrip patch antennas, in Handbook of Microstrip
Antennas, J. R. James and P. S. Hall, (eds.), London: Peter Peregrinus, 1989, pp. 393453.
17. J. D. Jackson, Classical Electrodynamics, New York: Wiley, 1975.
18. S. Y. Liao, Microwave Devices and Circuits, Englewood Cliffs, NJ: Prentice-Hall, 1985.
19. C. Vittoria, Microwave Properties of Magnetic Films, Singapore: World Scientific, 1993.
20. R. F. Harrington, Time Harmonic Electromagnetic Fields, New York: McGraw-Hill, 1961.
21. H. How C. Vittoria, Radiation modes in dielectric circular patch antennas, IEEE Trans. Microw. Theory Tech., MTT-42:
1939, 1994.
22. D. M. Pozar, Input impedance and mutual coupling of rectangular microstrip antennas, IEEE Trans. Antennas Propag.,
AP-30: 1191, 1982.
23. H. An, K. J. C. Nauwelaers, A. R. Van de Capelle, Broadband microstrip antenna design with the simplified real
frequency technique, IEEE Trans. Antennas Propag., AP-42: 129136, 1994.
24. R. Q. Lee, K. F. Lee, J. Bobinchak, Characteristics of a two layer electromagnetically coupled rectangular patch antenna,
Electron. Lett., 23: 10701072, 1987.
25. C. K. Aanandan, P. Mohanan, K. G. Nair, Broadband Gap-Coupled Microstrip Antenna, IEEE Trans. Antennas Propag.
AP-38: 15811586, 1990.
26. G. P. Gauthier, A. Courtay, G. M. Rebeiz, Microstrip antenna on synthesized low-dielectric constant substrates, IEEE
Trans. Antennas Propag., AP-45: 13101314, 1997.
27. M. Zheng, et al., Broadband microstrip patch antennas on micromachined silicon substrates, Electron. Lett., 34 (1):
34, 1998.
28. P. S. Hall, Probe compensation in thick microstrip patches, Electron. Lett., 23: 606607, 1987.
29. K. F. Lee, et al., Experimental and simulation studies of coaxially fed U-slot rectangular patch antenna, IEE Proc.
Microw. Antennas Propag., 144: 354358, 1997.
30. K. M. Luk, et al., Broadband microstrip patch antenna, Electronics Lett., 34 (15): 14421443, 1998.
31. S. Dey R. Mittra, A compact broadband microstrip antenna, Microw. Opt. Technol. Lett., 11 (6): 295297, 1996.
32. A. Sabban, A new broadband stacked two-layer microstrip antenna, IEEE AP-S Int. Symp. Dig., 6366, 1983.
33. P. S. Bhatnagar, et al., Experimental study on stacked triangular microstrip antennas, Electron. Lett., 22: 864865,
1986.
34. R. Q. Lee K. F. Lee, Experimental study of the two-layer electromagnetically coupled rectangular patch antenna, IEEE
Trans. Antennas Propag., AP-38: 12981302, 1990.
35. H. Y. Yang G. N. Alexopoulos, Gain enhancement methods for printed circuit antennas through multiple superstrates,
IEEE Trans. Antennas Propag., AP-35: 860863, 1987.
36. K. Hiraswa M. Haneishi, Analysis, Design, and Measurement of Small and Low-Profile Antennas, London: Artech
House, 1992, Ch. 5.
37. K. Wang Y. Lin, Small broadband rectangular microstrip antenna with chip-resistor loading, Electron. Lett., 33 (19):
15931594, 1997.
38. S. Dey R. Mitttra, Compact microstrip patch antenna, Microw. Opti. Technol. Lett., 13 (1): 1214, 1996.
MICROSTRIP ANTENNAS
35
39. H. Y. Yang N. G. Alexopoulos, Generation of nearly hemispherical and high gain azimuthal symmetric patterns with
printed circuit antennas, IEEE Trans. Antennas Propag., 35: 972977, 1987.
40. D. Cailleu, N. Haese, P. A. Rolland, Microstrip adaptice array antenna, Electron. Lett. 32 (14): 12461247, 1996.
41. R. J. Dinger, A planar version of a 4.0 GHz reactively steered adaptive array, IEEE Trans. Antennas Propag., 34:
427431, 1986.
HOTON HOW
ElectroMagnetic Applications
CARMINE VITTORIA
Northeastern University
MONOPOLE ANTENNAS
513
MONOPOLE ANTENNAS
This article describes the concepts underlying monopole antennas and their many applications, such as for broadcasting,
car radios, and more recently for cellular telephones. In its
simplest form, the monopole antenna above an infinite ground
plane shown in Fig. 1(a) can be considered as one-half of a
corresponding double-length center-fed linear dipole shown
for comparison in Fig. 1(b). The current distribution for a vertical monopole antenna of height h is assumed to be a standing wave of the following form, in terms of the feed point current I(0):
I(z ) =
I(0)
sin k(h z )
sin(kh)
(1)
;;;
z
z = h
I(z )
Ground
z = 0
(a)
z = h
I(z )
z = 0
z = h
(b)
Figure 1. A vertical monopole antenna above ground (a) and the corresponding center-fed dipole (b). Shown also are the current variations I(z) over the lengths of both the monopole and the corresponding dipole.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
514
MONOPOLE ANTENNAS
Table 1. Relationships Between Monopole and Dipole Antennas
Monopole Above Ground
Length h
Radiation pattern
Corresponding Dipole of
Twice Length L 2h
Feed-point resistance Ra
Feed-point reactance Xa
Xa
Directivity Da
Da
monopole
Xa,d (2h)
2Da,d (2h)
monopole
monopole
ground are also related to those for the corresponding doublelength dipole in a fairly simple manner.
We give in Table 1 the relationships between a monopole
antenna of length h above ground and the corresponding dipole antenna of length L 2h. A graph of the variation of the
feed-point resistance and reactance of the monopole antenna
above ground is given in Fig. 2 as a function of length h/ ,
where is the free-space wavelength at the radiation frequency (1). Note that the reactance Xa depends on the conductor radius a, whereas the feed-point resistance Ra is relatively
independent of conductor radius a for thin antennas (a/
1). The radiation patterns of a few typical monopole antennas
are given in Fig. 3. Because a monopole antenna is symmetric
with respect to angle in the azimuthal plane (xy plane),
these patterns are independent of angle and depend only
on angle (see Fig. 1). The following relationship is given for
power density S():
2
;;;
;;;
;;;
;;;
;;;
(3)
(a)
S( ) =
30 W P ( )
r2 Ra sin2 (kh)
2
100
(2)
70
(b)
100
200
a
= 0.0035
60
0.0014
50
0.00035
0.00014
300
40
30
0.000035
400
20
500
10
600
0
0
0.10 0.15 0.20 0.25 0.30
Monopole length h/ (wavelengths)
Resistance Ra ()
Reactance Xa ()
(c)
(d)
Figure 3. Radiation patterns for some typical end-fed vertical monopoles above infinite ground: (a) h/ l 0.05 (short monopole), (b)
h/ 0.25 (quarter-wave monopole), (c) h/ 0.5 (half-wave monopole), and (d) h/ 0.75 ( monopole) [adapted from E. C. Jordan
and K. G. Balmain (1)].
MONOPOLE ANTENNAS
Ra
Ra + Ro
Insulator
;;
;;;
;;;;;
Support
tower
Vertical
radiator
Feed wire
Support
tower
Insulator
Ground
Equation (2) gives the power density radiated from a monopole antenna for a ground of infinite conductivity and for flat
earth. For finite conductivity ground, the power density at the
surface of the earth diminishes more rapidly than the square
of the distance r from the antenna (2). Furthermore, the rate
at which the power density at the surface of the ground diminishes with distance is steeper at higher frequencies. The
rate of reduction of the radiated power density as compared
to Eq. (2) is also higher for lower conductivity soil such as
sandy or rocky lands, cities, etc., rather than propagation over
seawater, marshy, or pastoral lands, etc.
For distances in excess of about 1000 miles, there is a precipitous drop in the surface power density at all frequencies
due to the increased effect of the curvature of the earth (2).
Monopole antennas are also the antennas of choice for very
low frequency (VLF) (3 to 30 kHz) and low frequency (LF) (30
to 300 kHz) communication systems. For these applications,
the height h of the antenna is generally a small fraction of
the wavelength. From Fig. 2, for small values of h/ , the feedpoint resistance Ra is very small, on the order of a few ohms,
and the ohmic losses resulting from the surface resistance of
the antenna can be significant by comparison. This results in
reduced antenna radiation efficiency r given by the following
equation:
r =
Flat top
Insulator
515
(4)
;;
;;
;;
;;
Ground plane
or reflector
(a)
(b)
(c)
Figure 5. Some typical geometries used for top-loaded monopole antennas: (a) plate, (b) umbrella of wires, and (c) inverted L monopole.
516
;;
;;
;;
MONOPOLE ANTENNAS
Ground plane
or reflector
(a)
Z0
(b)
Reflector
Driven antenna
Directors
(c)
(5)
where Ra and Xa plotted in Fig. 2 are the resistance and reactance of the single arm (not-folded) monopole antenna, and
the factor c is given approximately by
c
ln(d/a1 )
ln(d/a2 )
(6)
In Eq. (6), a1 and a2 are radii of wires or rods used for arms 1
and 2, respectively, and d is the center-to-center spacing between the two arms of the antenna, generally much smaller
than the height of each of the arms. For a folded monopole
antenna where equal radii arms are used, the feed-point impedance is, therefore, four times higher than that for a monopole antenna that is not folded. This can be truly advantageous, because the feed-point resistance may now be
comparable to the characteristic impedance Z0 of the transmission or feeder line, and the reactance of the antenna may
easily be compensated by using a lumped element with a reactance that is negative of the reactance (1 c)2 Xa at the
terminals of the folded monopole antenna.
Shunt-Fed Monopoles. A typical arrangement of a shunt-fed
grounded monopole is shown in Fig. 6(b). This corresponds to
one-half of a delta match that is often used for dipoles. Because the base of the monopole is grounded for this arrangement, the antenna is fed typically 15 to 20% farther up on the
antenna; the exact location is determined by matching the
Guys
o
To
transmitter
Ground counterpoise
Figure 7. A conical monopole.
Guys
MONOPOLE ANTENNAS
The rigorous approach is based on a series expansion of transverse magnetic (TM) modes and is computationally complex.
Numerical methods, based on the assumption that the antenna is built from a series of wires, have been commonly
used for specific monopole arrangements.
An example of a conical monopole constructed of 60 evenly
spaced wires above a ground plane is an antenna that has
been in use by the U.S. Navy (7). For use in the HF band 2
to 14 MHz, the dimensions of the monopole are l 30.6 m,
a 0.001 m, 0 45 (see Fig. 7). The radius of the ground
screen is approximately 37.5 m, constructed of 120 radial
wires over average soil (r 15 and 0.012 S/m), and the
radius of the screen wire is 0.002 m. This antenna performs
over a 7 : 1 frequency band (2 to 14 MHz) with an input resistance varying from 44 to 64 (6). Because of the fairly narrow range of variation of input resistance, matching of this
antenna to a 50 line is possible over at least a 7 : 1 frequency range.
Monopole Antennas for Car Radio and Mobile Communications. Monopole antennas used for car radio and mobile communications commonly rely on the body of the vehicle to provide the ground plane. Variations in the shape of the vehicle
and placement of the antenna affect the radiation pattern and
performance of these antennas (8). For an antenna mounted
on the automobile, it would no longer radiate isotropically in
the horizontal plane but may radiate with a slightly higher
gain in directions where the automobile body extends farther,
thus simulating a larger width ground plane in such directions.
For communication antennas in general, it is often desirable to have the shortest possible antenna. Although a quarter-wave monopole is easier to match, a shorter antenna has
an associated capacitive reactance. This can, however, be canceled out by adding a loading coil (inductance) to the base of
the antenna.
Another arrangement often used is to place an inductive
coil at the center of the antenna (9). Because of the altered
current distribution along the length of the antenna (similar
to a top-loaded monopole), this increases the feed-point impedance of the end-fed antenna to roughly twice the value,
making it easier to match to such an antenna. The value of
the inductance needed at the center is approximately double
that which would be needed at the ends. This consequently
increases the coil resistance.
In order to reduce the inductance of the loading coil required to match the antenna at either the center or top, a
capacitive load (a small metal ball) may be added to the tip
of the antenna.
In all this discussion of monopoles, we have assumed that
the wave propagation from the antenna is over a perfectly
conducting or infinite conductivity ground plane. This is
hardly ever the case. For ground wave propagation ( 90)
over a finite conductivity ground plane, the fields drop off
more rapidly than 1/r2 given by Eq. (2). This effect is more
pronounced for higher frequencies and is generally small only
at very low frequencies less than a few tens of kilohertz. Also,
as expected, the departure from the idealized Eq. (2) is more
pronounced the lower the conductivity of the soil. Rocky and
jagged hilly ground, for example, would result in a more rapid
drop off of the power density S along the ground plane than
propagation over sea water.
517
518
MONOPOLE ANTENNAS
3
3.6
10.2
16.8
Medical Applications
23.4
BIBLIOGRAPHY
1. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, 2nd ed., Englewood Cliffs, NJ: Prentice-Hall,
1968, p. 548.
2. ITT Handbook, Reference Data for Radio Engineers, 5th ed., Indianapolis, IN: Howard W. Sams & Co., Inc., Chap. 26, 1973.
(a)
3
3.6
10.2
16.8
23.4
Head model
O. P. GANDHI
G. LAZZI
C. M. FURSE
University of Utah
519
MULTIBEAM ANTENNAS
553
MULTIBEAM ANTENNAS
Multiple beam, or multibeam, antennas are systems designed
to produce several antenna beams from a single or multiple
apertures. Traditional antennas use a single focusing device,
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
554
MULTIBEAM ANTENNAS
ferent input port produces an antenna beam pointing in a different direction. The collection of input ports will produce
beams that span the antennas field of view. This field of view
is usually a rectangular sector of space. This type of MBA, as
with the reflector and lens antennas, has a scanning loss that
must be tolerated.
Feed horns are pyramidal, or conical, extensions of the
waveguide used to excite them. Their apertures are usually
packed together to form an equilateral triangular grid. The
centers of their apertures are spaced to produce the desired
angular spacing between adjacent beams. There is one horn
for each beam and each horn has at least one input port.
These input ports are connected to the output ports of a BSN
or BFN. In some cases the antenna system is dual-polarized
and each feed horn has two input ports, one for each polarization. In this case a BSN or BFN is connected to each port, of
like polarization, of the feed horn array.
A BSN consists of a corporate tree arrangement of singlepole double-throw switches at each junction of the corporate
tree. There is a single-input port and usually 2m (where m is
an even integer) output ports.
This circuit permits the input port to be connected to any
one of the output ports. A BFN consists of a corporate tree
arrangement of variable power dividers and phase shifters.
There is a single-input port and usually 2m output ports. This
circuit permits the input signal to be divided among the output ports in accordance with any desired amplitude and
phase distribution. Thus a BFN permits one to produce any
beam shape that the converging device is capable of producing. This includes a pencil beam pointing anywhere in the
antennas field of view, or a single beam that covers the entire
field of view.
A control subsystem consists of electronic equipment capable of interpreting input signals and converting them into
control signals that operate the switches of a BSN, or the
power dividers and phase shifters of a BFN. The input signals
may be in the form of voltages and the addresses to which
they must be supplied, or in the form of beam pointing angles.
The actual form is dictated principally by the system within
which the MBA must operate. It is possible to include auxiliary equipment, in the control subsystem, that will detect the
presence of undesirable signals and reduce their amplitude
by shaping the antenna pattern.
GENERAL THEORY
The general theory and design procedures are similar if either
a reflector or a lens is used as the focusing device. The following discussion uses a lens system as a model and assumes
that it is transmitting. A general configuration is shown in
Fig. 1. Exciting the feed horn labeled #1 produces an antenna
beam pointing along the focal axis of the lens. Exciting the
feed horn labeled #2 produces a beam pointing in the direction , measured from the focal axis. Exciting the feed horn
labeled #3 produces a beam pointing in the direction , also
measured from the focal axis. Exciting additional feed horns,
located in the focal region of the lens, will produce additional
antenna beams. Usually the focal length of the lens is approximately equal to its diameter and the angle is approximately equal to . Usually the feed horns are arranged on
an equilateral triangular grid; consequently the centers of the
MULTIBEAM ANTENNAS
Lens
d
555
Nb = int[(/ )2 ]
(2)
2
1
3
1
2
Focal axis
(3)
Typical MBA
Figure 1. Typical MBA.
(1)
Coverage area
Field of view
(4)
(5)
(6)
556
MULTICAST ALGORITHMS
Multiple-Aperture MBA
BIBLIOGRAPHY
A. R. Dion and L. J. Ricardi, A variable-coverage satellite antenna
system, Proc. IEEE, 59: 252262, 1971.
J. W. Duncan, S. J. Hamada, and P. G. Ingerson, Dual polarization
multiple beam antenna for frequency reuse satellites, AIAA/CASI
6th Commun. Satellite Systems Conf., Montreal, Canada, 1976.
IEEE Definitions of Terms for Antennas, IEEE Standards 45, 1973.
J. T. Mayhan and L. J. Ricardi, Physical limitations of interference
reduction by antenna pattern shaping, IEEE Trans. Antennas Propag., AP-23, 1975.
J. T. Mayhan, Nulling limitations for a multiple-beam antenna, IEEE
Trans. Antennas Propag., AP-24, 1976.
B. M. Potts, Radiation pattern calculations for a waveguide lens multiple-beam antenna operating in the AJ mode, MIT, Lincoln Laboratory, Technical Note 1975-25, 1976.
L. J. Ricardi et al., Some characteristics of a communication satellite
multiple-beam antenna, Defense Tech. Inf. Center AD/A-006405,
1975.
L. J. Ricardi, Methodology of assessing antenna performance, Lincoln
Laboratory Technical Note 1978-24, 1978.
LEON J. RICARDI
Creative Engineering
RADAR ANTENNAS
Although the principle of radar has been known and advocated by many scientists since the early 1900s, it
did not come into its own until its widespread development and application in World War II when the word
RADAR was coined. Since then, the use of radar has expanded prodigiously. Important applications are found
not only in numerous military problems but also in many public and commercial areas (e.g., civil air-traffic
control, aircraft navigation, ship safety, spacecraft, ground detection, and remote sensing of the environment).
It is even applied to law enforcement when police use radar to check the speed of cars.
The major types of radar are pulse radar, high-range-resolution radar, continuous waveform (CW) radar,
moving target indication (MTI) radar, airborne radar, tracking radar, pulse Doppler radar, ground penetrating
radar, synthetic aperture radar (SAR), and inverse SAR. Each of these radar types employs a different characteristic waveform, which is determined by the corresponding radar antenna and signal-processing method.
The radar antenna is an important component of radar. For different radar types, the requirement for
antenna performances and functions is varied. Because the subject of radar antennas is very extensive, only
selected types and performances of radar antennas are discussed in this article. The reader should consult the
bibliography for further details (1,2,3,4,5).
Radar antennas serve as transduces to couple electromagnetic (EM) energy into the atmosphere. A simple
description of the antenna in a radar system is shown in Fig. 1. The functions of the antenna when transmitting
are to concentrate the radiated energy from the transmitter into a shaped directive beam in a predetermined
direction. When receiving, the antenna again forms a beam in a particular direction and collects the energy
contained in the reflected target echo signals. Received energy is sent to the receiver via transmission lines.
Therefore, the primary purpose of the antenna is to determine accurately the angular direction of the target
in both transmit and receive modes. For this purpose, a highly directive beamwidth is required, not only to
achieve accurate angular targets but also to resolve closely located targets. This important feature of radar
antennas can be quantitatively expressed in terms of the beamwidth, gain, and sidelobes.
Generally, the functions of radar antennas can be classified into beam scanning and target tracking. Beam
scanning fulfills the search or surveillance function of a radar. It requires that the narrow directive beam be
scanned rapidly and repeatedly over the region where the detected target may appear. Target tracking, however,
requires that the narrow beam follow the target once it has been detected. In some radar systems, particularly
airborne radars, the antenna is designed to perform both searching and tracking functions.
The preceding functional description of radar antennas implies that a single antenna is used for both
transmitting and receiving via the duplexer, a device that permits both transmission and reception of EM
waves with the single antenna. This is true for most radar systems, but there are exceptions (e.g., the bistatic
radars in which separate transmit and receive antennas are used). This article considers only the single
antenna.
Radar antennas are mainly classified into two types: quasi-optical antennas, which can be analyzed by
optics methods like geometrical optics and physical optics, and array antennas. In some cases, small antennas
are used (e.g., in ground-penetrating radars). Quasi-optical antennas include reflector antennas and lens
antennas. The former are still widely used for radar, whereas lens antennas, although still used in some
1
RADAR ANTENNAS
Fig. 1. Basic elements in a radar system in which the antenna is an important part. It can be served as transmitting and
receiving.
communication and military applications, are no longer used in modern radar systems. Therefore, because of
limited space, we discuss only reflector antennas and array antennas.
where P0 is the total power consumed by the antenna, and Pr is the power radiated by the antenna. The later
can be expressed by the radiation intensity (, )
in which is the azimuth angle and is the elevation angle, as shown in Fig. 2. From Eq. (2), the average
radiation intensity is easily obtained
From Eqs. (1)(3), we can define some basic parameters of an antenna. Directivity is a measure of the
ability of the antenna to concentrate the radiated power in a particular direction. Thus it can be defined as the
RADAR ANTENNAS
Fig. 2. Spherical coordinate system. The azimuth angle and elevation angle are adopted by most antennas.
In practice, one is usually interested in the maximum directivity of the main lobe.
Gain is another important parameter of an antenna. It represents the ability to concentrate the power
accepted by the antenna in a particular direction:
Thus antenna gain is always less than directivity except for a lossless and reflectionless antenna (= 1). Again,
the peak value of the gain, G0 , is of more interest in practice.
The aperture of an antenna is its physical area projected on a plane perpendicular to the desired direction.
If the antenna is lossless and the aperture of area A is uniformly illuminated with equiphase, the directivity is
given by
which is the maximum available gain from an aperture A. Practical antennas are not uniformly illuminated,
but they have a maximum in the center of the aperture and are less tapered toward the edges in order to reduce
the sidelobes of the pattern. In this case, an effective aperture is defined from the directivity
The concept of effective aperture is very useful when considering the antenna in its receiving mode, which
measures the effective absorption area of the antenna to an incident plane wave. Let Ae be the peak value of
RADAR ANTENNAS
Ae (, ). Clearly, the effective aperture Ae is always less than the physical aperture A by a factor a
which is usually called the aperture efficiency. From the definitions in Eqs. (6)(8), we can see that the aperture
efficiency measures only how effectively a given aperture is used, but does not involve the EM energy loss.
Therefore, the efficiency of the antenna should be the multiplication of a and antenna losses L .
When the power radiation intensity (, ), the directivity D(, ), the gain G(, ), and the effective
aperture Ae (, ) are normalized to their peak values, they will be identical and are called the antenna radiation
pattern, which in fact represents the EM-energy distribution in three-dimensional (3-D) angular space. There
are several ways to plot the radiation pattern: rectangular or polar, voltage intensity or power density, absolute
value or dB value, or power per unit solid angle. However, the 3-D plots of radiation pattern require extensive
data. Thus two-dimensional (2-D) plots are usually adopted as a result of the ease in measurement and plotting.
Generally, the two cuts of 3-D plots in the principal elevation and azimuth planes are sufficient to describe the
pattern performance of an antenna, which is much less costly.
Instead of azimuth and elevation, E- and H-plane patterns are usually used in actual radar antennas
because the terms azimuth and elevation imply the earth-based reference coordinates, which are not applicable
to some space-based systems like airborne and satellite.
The main lobe of the radiation pattern is in the direction of maximum gain; all other lobes are called
sidelobes. In addition to the peak gain of the main lobe, two other important features of an antenna pattern are
the beamwidth of the main lobe, which is usually specified at the half-power level (3 dB), and the maximum
sidelobe level. The half-power beamwidth (HPBW) is usually a measure of the resolution of an antenna.
Therefore, if two identical targets at the same range can be separated by the HPBW, they are said to be
resolved in angle.
The beamwidth of an antenna is determined by the size of the antenna aperture as well as the amplitude
and phase distributions across the aperture. For a given distribution, the half-power beamwidth in a particular
plane is inversely proportional to the size of the aperture in that plane
where L is the aperture dimension and K is a constant for the given distribution, which is known as the
beamwidth factor. Accurate estimate of the beamwidth factor must take into account the aperture illumination function. However, K = 70 can give a rough estimate in most cases. For reflector-type antennas, a good
estimation to K is given by (6)
where C is a unitless constant and 1 and 2 are HPBWs in degrees. The accurate value of C depends on the
antenna efficiency, but the rough estimate can be taken from 26,000 to 35,000 (1,7).
RADAR ANTENNAS
Fig. 3. Common reflector antenna types. (a) Paraboloid; (b) parabolic cylinder; (c) shaped reflector; (d) multiple beam; (e)
monopulse; (f) cassegrain.
Ideally, an antenna radiation pattern would consist of a single main lobe and no sidelobes. However, for all
practical radar antennas, it contains numerous sidelobes. These sidelobes can be a source of problems for the
radar system. For example, a radar for detecting low-flying aircraft targets can receive strong ground or ocean
echoes, which is called the clutter, through the sidelobes. This signal will interfere with the desired echoes
coming from the low radar cross-section targets through the main lobe. Another fatal problem of the sidelobes
is coming from jamming, which threatens most military radars. Therefore, it is often (but not always) desirable
to design radar antennas with sidelobes as low as possible to minimize such problems. However, low sidelobes
and high gain are competing requirements. Thus, the trade-off between sidelobes level, gain, and beamwidth
is an important consideration for designing radar antennas.
RADAR ANTENNAS
literature (1,5). In this antenna, the feed is placed at the focus of the paraboloid. Then the spherical wave
emerging from the feed strikes on the reflector and is transformed into a plane wave after reflection to form
a pencil beam. Usually, the paraboloidal antenna provides a high gain and a minimum beamwidth with the
simplest and smallest feed.
An important parameter for this paraboloid is the ratio of the focal length f to the aperture diameter D.
For fixed aperture, when f increases, the reflector becomes flatter and introduces less distortion for polarization
and off-axis beam. However, a narrow primary beam that leads to a larger feed is also required. For example,
the size of the horn to feed a reflector of f /D = 1.0 is approximately four times that of f /D = 0.25. In general,
most reflectors are chosen to have a ratio between 0.25 and 0.5.
Besides the round aperture, a variety of reflector outlines are also used in practice (2,5). For example, an
oblong shape is chosen when the azimuth and elevation beamwidths have different requirements.
Parabolic-Cylinder Antennas. It is very common that only one of the azimuth and elevation beams must
be steerable or shaped while the other is not. In this case, a parabolic-cylinder reflector fed by a line source
shown in Fig. 3(b) can meet the requirement at a modest cost. The line source feed may be a parallel-plate lens,
a slotted waveguide, a phased array, etc. (1). Of course, the parabolic-cylinder antenna is not limited to these
applications. It can even be applied when both patterns are fixed in shape (e.g., the AN/TPS-63).
Parabolic cylinders suffer from large blockage if they are symmetrical, thus they are often built offset. Properly designed, however, a cylinder fed by an offset multiple-element line source can have excellent
performance (8).
Shaped-Reector Antennas. A good method of shaping the beam in one plane is using shaped reflectors
(9), as shown in Fig. 3(c), in which the surface itself is no longer a paraboloid. From Fig. 3(c), each portion of the
reflector is aimed in a different direction, making the analysis complicated. However, with modern computers,
arbitrary beam shapes can be approximated accurately by direct integration of the reflected primary pattern.
A disadvantage of the shaped reflector is that its effective aperture is very small because a large fraction
of its energy is radiated to a wide angle. However, it can reduce the blockage from the feed. In most shaped
reflectors, the feed is usually placed outside the secondary beam. Then the blockage can be virtually eliminated
even though the feed appears in front of the reflector, as shown in Fig. 3(c).
Comparing shaped-reflector antennas with parabolic-cylinder antennas, the shaped reflector is cheap and
simple to construct. But it has less control over the beam shape because only the phase of the wave across the
aperture is changed, whereas both phase and amplitude can be adjusted in the linear array for the parabolic
cylinder.
Multiple-Beam Reector Antennas. Very often the radar designer needs multiple beams to provide
extended coverage. This requirement can be achieved by using a reflector with multiple feeds (10), as shown
in Fig. 3(d). As we know, a feed at the focal point of a paraboloid forms a beam at the direction parallel to the
focal axis. The feeds displaced from the focal point, however, will form additional beams at distinct angles from
the axis with nearly full gain. Therefore, the coverage range of the radar can be increased. However, when the
feed is off the focus, a distortion occurs on the antenna pattern. As the angular displacement increases, the
beamwidth becomes larger, and additional sidelobes arise. The other limitation of this kind of antennas is that
the extended feeds increase the blockage of the aperture.
Monopulse Reector Antennas. An especially common multiple-beam radar antenna is the monopulse
reflector illustrated in Fig. 3(e). In fact, the illustration shows an amplitude comparison system, which is far
more prevalent than the phase comparison system. In the amplitude system, the sum of the even-numbered
source outputs fed by a single pulse forms a high-gain and low-sidelobe beam, and the difference forms a
precise deep null at the peak-value direction of sum pattern. The sum beam is used on bolt transmit and
receive to detect the target, whereas the difference beams provide angle determination in both azimuth and
elevation angles. Therefore, the monopulse reflector is normally used in tracking systems in which the antenna
is removable and keeps the target near the null in difference pattern.
RADAR ANTENNAS
Fig. 4. Offset dual reflector. The blockage of the subreflector can be virtually eliminated.
A trade-off to design the monopulse radar antenna is the choice of feed size. The conflict arises between
the goals of high sum-beam gain and large difference-beam slopes. For example, for a four-horn feed, the former
requires a small overall horn size to reduce the blockage, whereas the latter requires large individual horns.
Also, high difference sidelobes are required for tracking purpose in the design. To overcome these problems,
many methods have been presented and several suggested configurations of the feed are given in Ref. (11).
Multiple-Reector Antennas. Multiple-reflector systems can offer more degrees of flexibility by shaping
the primary beam and then overcoming some of the shortcomings of single reflector. Figure 3(f) shows the most
common type of multiple reflector, the Cassegrain antenna (12), in which the feed system is conveniently located
behind the main reflector. In the Cassegrain system, the subreflector is a hyperboloid. The feed is placed at the
left focus of the hyperboloid, and the paraboloid focus coincides with the right focus. From geometry optics, the
spherical wave radiated by the feed and then reflected to the main reflector by the hyperboloid is equivalent to
the one emerging directly from the right focus. Therefore, the wave reflected from the main reflector is also a
plane wave that forms a pencil beam. A similar antenna to Cassegrain is the Gregorian antenna, which uses
an ellipsoidal subreflector rather than a hyperboloidal one.
Aperture blockage can be very large for symmetrical Cassegrain antennas, as illustrated in Fig. 3(f). There
are two ways to reduce blockage. The first method is to make the diameter of the subreflector equal to that of
the feed, but the reduction is limited. The second method, which is generally used, is achieved by offsetting
both the feed and subreflector. For example, Fig. 4 shows an offset dual reflector with blockage and supporting
struts virtually eliminated. This antenna usually has very low sidelobes.
The geometry of the Cassegrain antenna is especially attractive for monopulse tracking radar because
the radiofrequency (RF) plumbing can be placed behind the reflector to avoid blocking of the aperture. Also, the
long runs of transmission line out to the feed at the focus of a conventional monopulse reflector are avoided.
Therefore, the Cassegrain antennas have usually a low noise temperature.
Feeds for Reector Antennas. In the previously discussed antennas, a spherical wave from the focus
of the paraboloidal reflectors will be transferred into a plane wave to form a pencil beam at the transmit mode,
while the incoming plane wave will be converted into spherical phase fronts centered at the focus on their
receive mode. Thus the feeds for reflector antennas must be point-source radiators so that they can radiate
spherical waves. Other requirements for reflector feeds include the proper illumination of the reflector with a
prescribed amplitude distribution, a minimum spillover, and the correct polarization with no or minimum cross
polarization. These are the basic factors chosen to design a feed. For a different application of the antennas,
other considerations must be included.
RADAR ANTENNAS
The most common feeds for reflector antennas are the flared waveguide horns, which are very suitable
for microwave-frequency radar systems (L band and higher). At lower frequencies, dipole feeds, Yagi-Uda or
log-periodic feeds are sometimes used. For example, a linear array of dipoles can feed a parabolic-cylinder
reflector. Rectangular waveguide horns propagating the dominant TE01 mode are widely adopted because they
meet the high power and other requirements, although in some cases circular waveguide horns propagating
TE11 mode have also been used. These single-mode horns provide just one linear polarization. When such
additional requirments as polarization diversity, multiple beams, high beam efficiency, or ultralow sidelobes
are needed, the design for feeds becomes more complicated. In these cases, segmented horns, finned horns,
multimode horns, and corrugated horns must be used. Figure 5 illustrates some of the types, which have been
well studied (1,13).
In addition, the sitting of the feeds has also a great influence to the performance of the reflector antennas.
Figure 6 shows several placement styles of the feeds.
Reector Antennas Design and Analysis. The analysis and design of conventional paraboloidal
reflectors are well developed (1,5) and not stated here. However, most of the rotating search radars require
that the antenna patterns have a narrow azimuth beamwidth for angular resolution and a shaped elevation
beam for multiple requirements, for example, the ARSR-3 radar for air-traffic control and the AN/TPN-19 for
ground control. Thus, a shaped reflector is nearly always the practical choice. In this section, we will consider
the analysis and design of such kinds of antennas.
Requirements. When a search radar works, small nearby targets like birds and insects may clutter the
output of the radar because of the inverse fourth-power variation of signal strength with range. For example,
the radar that can detect a 1 m2 target at 100 nmi can detect a 10 4 m2 target at 10 nmi. [Nautical mile (nmi)
where 1 nmi = 1852 m is usually used as the range unit.] To avoid the clutter, the receiver gain should be
reduced at short range and increased at distant range so that the received signal from a target of constant size
remains unchanged with range. The programmed control of the receiver gain to keep a constant echo signal
strength is called sensitive time control (STC), which is an effective method of eliminating the radar echoes
from unwanted objects. Therefore, STC is the usual requirement for designing the search radar.
A typical range coverage requirement for the rotating search radars is shown in Fig. 7, which is an
idealized simplification of the STC requirement (14). It gives a relationship between the radar range, height,
RADAR ANTENNAS
and elevation angle. From Fig. 7, at low elevation angles (0.5 to 3.5 ), the maximum range is the critical
requirement. When the elevation angle increases (3.5 to 9.5 ), however, the height becomes the governing
requirement. In this period, the corresponding pattern is a cosecant square. At high elevation angles (9.5 to
30 ), the STC is used.
The two-way coverage requirement shown in Fig. 7 is inconvenient for antenna design. It can be converted
into a pattern requirement if we plot only the range against angle because the range is directly proportional
to the pattern amplitude, as shown in Fig. 8 (the solid line).
Analysis. Antenna analysis is very important not only in analyzing a given antenna but also in antenna
design. Because the accurate design is usually an iterative procedure, in each iterative cycle, an analysis is
needed.
There are two methods to analyze a given antenna: current-distribution method and aperture-field method
(5). For shaped antennas, however, the first one is more suitable. From the magnetic field H radiated by the
feed, one can get the induced current on the reflector surface J = 2I H. Then the radiated fields by the
reflector can be obtained by integration.
This type of accurate analysis is quite time consuming. In practical design, the antenna pattern and
gain are usually approximated in the first several iterative cycles. For rotating search radars, the antenna
pattern can be obtained directly from the coverage requirement; however, the designer must realize that the
range-angle plot represents the minimum range requirement. If some ripple is anticipated (normally about 1
dB), the design curve should be raised 1 dB in the height limit and STC coverage regions. For the range limit
region, a reasonable situation is achieved by assuming that a section or several sections of the reflector are
shaped so that the 3 dB points of the resulting pattern will coincide with the two corners of the range limit
portion, as shown in Fig. 8 (dashed line), which is just the realizable elevation pattern of the antenna.
For the gain of shaped beam considered here, we assume that the azimuth and elevation power patterns
are separable. Then using the definition in Eq. (5), the maximum gain (in main beam direction) can be expressed
10
RADAR ANTENNAS
Fig. 7. Typical two-way coverage requirement of the search radar, in which a relationship between the radar range,
height, and elevation angle is given.
as
where f az () and f el () are azimuth and elevation power patterns. For simplicity, the azimuth power pattern
can be approximated by a Gaussian function: f az () = exp[2.7726(/BWAZ)2 ]. Then
in which BWAZ is the azimuth half-power beamwidth. For the other integral in Fig. 8.
Design. For shaped-beam antennas, the shaping is generally in one plane (elevation), with a narrow
pattern of conventional design in the orthogonal plane (azimuth). The parabolic-cylinder antenna fed from a
line source is convenient for obtaining independent control of the patterns. However, it is much more expensive
and heavier than a reflector fed from a point source. In order to obtain a shaped beam, the antenna with a
point-source feed requires a reflector surface with double curvature. An easy but efficient way to do this is to
use an offset paraboloidal antenna by adding several shaped sections. Each successive section points slightly
higher than the preceding section, as shown in Fig. 9. Clearly, the projections of added sections are virtually
missing the feed. So there is no feed blockage. The design of the added sections can be done iteratively by using
a computer. However, the antennas width and height can be directly approximated from the antenna patterns
(14)
RADAR ANTENNAS
11
Fig. 8. Coverage requirement (solid line) and realizable pattern (dashed line).
in which BWEL is the elevation half-power beamwidth, and is the integral in the range-limit region.
Practical Consideration. In the analysis and design of radar antennas, some practical considerations
must be taken into account because they can affect the performances of the antennas. There are a lot of
practical details for different radar antennas, but here we consider only four common effects: feed-support
blockage, surface leakage, surface tolerance, and radomes.
Feed-Support Blockage. No matter whether the feed is placed in the front or rear, there is always
blockage from the feed supports. In the case of Cassegrain or Gregorian, the blockage comes from the support
of the subreflector. Like the blockage of feeds, the supports have two effects on the antennas performance: they
put a shadow in the desired distribution, and they scatter the intercepted power. Both effects can reduce the
gain and increase the sidelobe level.
The effective size of the blockage may be larger than its projected area depending upon the material and
polarization. For conducting supports, the effective size equals the actual dimension in the E plane, whereas
(where W and d are the width and thickness of a strut) in the H
the effective width becomes W +
plane (15). Therefore, the blockage of a strut of 0.125 square in cross section will be three times larger when
it is parallel to the E field than when it is parallel to the H field. In addition, various ways of sitting the
supports have different effects. Figure 10 shows three common ways of supporting the feeds. Note that the
tripod arrangement produces sidelobe ridges that are broader but 6 dB lower than those of the quadrapods
(15). The feed-support blockage can be reduced by using a dielectric strut.
Surface Leakage. Many reflectors are designed with a tubular grid or a wire-mesh instead of a solidmetal surface in order to reduce the wind resistance and the antennas weight. To accomplish this, the opening
12
RADAR ANTENNAS
area in the structure should be as large as possible. However, the openings lead to surface leakage, from which
backlobe is produced. The larger the area of the opening is, the higher the backlobe is. Therefore, a suitable
design of the surface structure is important to the antennas performance. Figure 11 illustrates some of the
common reflector surfaces, in which the gap S between conductors must be much smaller than half wavelength
in the H plane so that the passage of EM energy will be below cutoff (14).
Surface Tolerance. Mechanical consideration to design reflector antennas is very important in a radar
system. In most cases, it is even more difficult and complicated than the electrical design, especially for a
high-performance radar. The surface tolerance is one important part of the mechanical design. The reflector
surface must be smooth enough so that it remains within close tolerances of the ideal surface. Typically, the
surface error must be less than /28.
The roughness of reflector surface can cause small phase error on the aperture. From the antenna tolerance
theory, the loss of gain due to phase error is approximated by Eq. (16):
RADAR ANTENNAS
13
Fig. 11. Common types of reflector surfaces that can reduce the wind resistance.
where G1 and G0 are gains with and without phase error, and 2 is the mean square phase error. From Eq. (16),
one clearly sees that for 1 dB loss of gain the phase error
must be less than 0.45 rad, which means the
reflector surface error must be less than /28.
Radomes. Antennas for ground-based radars are often subjected to high winds, icing, and temperature
extremes so they must be sheltered if they are to survive and perform under adverse weather conditions. The
shelter to protect antennas is called a radome. Antennas mounted on aircraft must also be housed within a
radome to offer protection from large aerodynamic loads. An ideal radome should be perfectly transparent to
the rf radiation from (or to) the antenna but mechanically strong if they are to provide the necessary protection.
These two requirements are usually competing, thus the design of radomes must be compromised (17).
The design of radomes for antennas can be divided into two separate classes, depending upon whether the
antenna is for airborne or ground-based (or ocean-based) applications. The airbone radome is characterized by
smaller size than ground-based radomes because the antennas that can be carried in an aircraft are generally
smaller. The airborne radome must be strong enough to form a part of the aircraft structure and usually it is
designed to conform to the aerodynamic shape of the aircraft, missile, or space vehicle in which it is to operate.
If the antenna scans inside the streamlined radome, the incident angle may be varied around the normal
direction. This will lead to great degradation of antenna performance.
A properly designed radome should distort the antenna pattern as little as possible. The presence of a
radome can affect the gain, beamwidth, sidelobe level, beam direction, the voltage standing-wave ratio (VSWR),
and antenna noise temperature. Sometimes in tracking radars, the change rate of the beam direction can be
very important.
14
RADAR ANTENNAS
where HPBW is the 3 dB beamwidth (in degree) of the antenna pattern. When the beam points in the broadside
direction to the aperture, the corresponding antenna gain is
where is the antenna efficiency, which accounts for antenna losses (L ) and reduction in gain caused by
weighting the elements with a nonuniform amplitude distribution (a ).
RADAR ANTENNAS
15
Fig. 12. General structure of a phased array in which the phase shifter driver provides different phases to the array.
When the antenna is scanning to an angle 0 , the scanned beamwidth is increased from the broadside
beamwidth
whereas the gain of the planar array is reduced to that of the projected aperture
These characteristics can provide an easy way to estimate quickly when analyzing and designing array
antennas.
Types of Arrays. The arrays used in radar systems can be classified by such different means as
geometry and functions. This subsection will discuss several typical arrays.
Linear Arrays. The linear array generates a fan beam when the phase relationships are such that the
radiation is perpendicular to the array. If the radiation is at some other angles, the antenna pattern is a
conical-shaped beam. The broadside linear-array antenna may be used where broad coverage in one plane and
a narrow beamwidth in the orthogonal plane are required. An active application of linear array is that it can
act as the feed for a parabolic-cylinder antenna, as discussed earlier.
Planar Arrays. The two-dimensional planar array is capable of steering the beams in two angular
coordinates. It is probably the array of most interest in radar applications because of its versatility. Generally,
a rectangular aperture can produce a fan-shaped beam, whereas a square or a circular aperture can generate a
pencil beam. The array may also be made to produce simultaneously many search and or tracking beams with
the same aperture. Most of arrays discussed later will be the planar arrays.
Conformal Arrays. Sometimes, a radar designer is required to place array elements on an arbitrary
surface to achieve a directive beam with good sidelobe level and efficiency, which can be easily scanned
16
RADAR ANTENNAS
electronically. For example, on an aircraft, arrays arranged along the nose, on the wings, or on the fuselage
would be attractive options. In this case, a conformal array, which conforms to the geometry of a nonplanar
surface, should be used (1).
In principle, the array on any surface can be made to radiate a beam in some given direction by applying
the proper phase, amplitude, and polarization at each element. In practice, however, it is very difficult to control
the beam shape and obtain low sidelobes from an arbitrary surface when the beam is electronically scanned.
Furthermore, the mechanisms for feeding the elements and generation of the phase-shifter command are also
more complicated than those of a planar array. Therefore, the application of conformal array is only possible
for some simple shapes like cylinder and cones.
Most of the work on conformal arrays has been with the cylinder because it has a geometry suitable for
antennas that scan 360 in azimuth. Although it is of a relatively simple shape compared with others, the
properties of the cylindrical array are not as suitable as those of planar array. In the cylindrical array, the
radiation pattern cannot be separated into an element factor and an array factor as they can in the planar
array. Considering the additional difficulties in the practical control, the conformal arrays are not widely used
in the radar systems.
Thinned Arrays. Most array antennas have equal spacings between adjacent elements. In order to
obtain a given beamwidth with considerably fewer elements, a thinned array has sometimes been considered.
Generally, the shape of the main beam has little distortion after thinning. However, the average sidelobes are
degraded in proportion to the number of elements removed. Figure 13 shows an example of thinned arrays, in
which 77.5% of the elements are randomly removed from a regular grid (19). Then the gain, which is caused by
the actual element number, will drop by 6.5 dB, while almost 77.5% of the power is delivered to the sidelobes
because the main beam is nearly unchanged.
Therefore, the design of a thinned array consists of selecting the size of the aperture to give the desired
beamwidth, selecting the number of actual elements to give the desired gain, and arranging the element
distribution to obtain some required properties of the sidelobes.
The thinned arrays have seen only limited applications in radar system because of the reduction in gain
and the increases in sidelobes. Instead of thinning elements, the number of phase shifters can also be thinned,
in which some of the phase shifters in the array can be used to adjust the phase of more than one element. A
50% saving of the phase shifters might be possible (20).
Phase-Scan Arrays. The beam of an antenna points to the direction perpendicular to the phase front.
In phase-scan arrays, the phase front is adjusted by controlling the phase of each radiating element to steer
the beam, as shown in Fig. 14(a). The phase shifters are electronically actuated to perform rapid scanning and
are adjusted in phase to a value between 0 and 2 rad. If the spacing between two adjacent elements is S, the
incremental phase shift will be = (2/)S sin 0 , in which 0 is the scan angle. Because the phase shifters
have phase shift that is virtually independent of frequency, the scan angle 0 is frequency-dependent.
Time-Delay-Scan Arrays. Phase scanning was seen to be frequency-sensitive. Time-delay scanning
provides a way that is independent of frequency. Instead of the phase shifters, delay lines are used in the
time-delay-scan arrays, as illustrated in Fig. 14(b). For the spacing distance between two adjacent elements S,
the incremental time delay from element to element is t = (S/c) sin 0 , in which c is the velocity of propagation
of wave and 0 is again the scan angle. The time-delay circuits accompanied by all radiating elements are
normally too cumbersome. A reasonable compromise solution is sharing one time-delay network by a group of
elements in which each element has its own phase shifter. However, this will cause a grating lobe problem.
Frequency-Scan Arrays. A change in frequency of an EM signal propagating along a transmission line
produces a change in phase. This provides a simple way for obtaining the electronic phase shift, as shown in
Fig. 14(c). In this case, the incremental phase shift is
RADAR ANTENNAS
17
Fig. 13. (a) Thinned array with a 4000-element grid containing 900 elements; (b) typical pattern for a thinned array in
which SA is the average sidelobe level. [From Willey (15, Figs. 5 and 6) courtesy of Bendix Radio Div., Bendix Corporation.]
in which l is the length of line connecting adjacent elements. When the beam points broadside (0 = 0), Eq. (21)
yields m = l/0 , where 0 is the wavelength corresponding to the beam position at broadside. Then the direction
of beam pointing can be obtained from Eq. (21)
Comparing with other ways, the frequency-scanning system is relatively inexpensive and easy to implement.
Frequency-scan arrays have been developed and used to provide elevation-angle scanning. Combined with the
mechanical horizontal rotation, it served as 3-D radars in the past.
18
RADAR ANTENNAS
Fig. 14. Scanning of arrays. (a) Phase scan; (b) time-delay scan; (c) frequency scan.
Multiple-Beam-Forming Arrays. One of the properties of the phased array is the ability to generate
multiple independent beams simultaneously from a single aperture. In principle, an N-element array can
generate N independent beams. Multiple beams allow parallel operation and can obtain a higher data rate
than a single beam. The multiple beams may be fixed in space, steered independently, or steered as a group. In
some applications, multiple beams are generated on receive and connected to separate receivers, whereas only
one wide radiation pattern is needed on transmit so that it can cover all the received beams. Such multibeam
systems have found application with mechanical rotation for 3-D coverage.
Low-Sidelobe Arrays. A low sidelobes level is one of the most important requirements for the radar
antennas, especially for military radars that may be threatened by the jamming. Therefore, the low sidelobe
level has been and will be of great interests to radar designers. By modern technology, an ultralow sidelobe has
been achieved in the Airborne Warning and Control System (AWACS) radar, which now supports sidelobe levels
of more than 50 dB below the main-beam peak (21). However, as mentioned earlier, the low sidelobe and high
gain are competing requirements. Thus the cost to achieve low sidelobe is a reduction in gain and an increase in
beamwidth. Considering the practical implementation, the low sidelobes also require a high-tolerance control
and a good environment free from obstructions that could increase the sidelobes so low-sidelobe antennas are
usually expensive. In spite of these drawbacks, the trend to low-sidelobe arrays has accelerated because low
sidelobes provide an excellent deterrent to electronic countermeasures (ECM).
From the aperture theory, antenna sidelobes are related to the aperture amplitude distribution. For
phased arrays, the amplitude taper is generally fixed and does not change. However, the amplitude of each
radiating element can be controlled individually by the computer, which makes it possible to achieve very low
sidelobes.
RADAR ANTENNAS
19
The relation between the antenna pattern and the aperture illumination has been studied extensively
in many literatures. It has been shown that the far-field pattern is just the Fourier transform of the aperture
distribution. Some typical illumination functions and their corresponding gains, sidelobes, and beamwidths
can be found in Refs. 5 and 18. For low-sidelobe arrays, the Taylor illumination for the sum patterns and
the Bayliss illumination, a derivative form of the Taylor illumination, for the difference patterns are usually
chosen because they can provide low sidelobes at a minimum loss in gain. Note that the sidelobes predicted by
the aperture illuminations are suitable for perfect phase and amplitude distributions (error free). In practice,
aperture illuminations must be chosen to provide peak sidelobes below the requirements to allow for errors.
The effect of errors on the design will be mentioned next.
Practical Considerations. The array theory gives only the theoretical analysis and design of the
phased arrays. For practical use in radar systems, a lot of details must be considered. In this subsection, some
important considerations are addressed.
Phase Shifters. As already stated, there are three basic techniques to steer the beam electronically:
frequency scanning, time-delay scanning, and phase scanning with phase shifters. In practice, the use of phase
shifters is the most popular technique, but many other techniques have been developed for a variety of phase
shifters (22). Generally, the phase shifters can be classified into two categories: reciprocal and nonreciprocal.
For reciprocal phase shifters, the phase change does not depend on the direction of propagation. Therefore, it
is not needed to switch the phase states between transmit and receive if the reciprocal phase shifters are used.
For a nonreciprocal phase shifter, however, it is necessary to change the phase states between transmit and
receive. Typically, it takes a few microseconds to switch the nonreciprocal phase shifters. During this period,
the radar cannot detect targets.
A perfect phase shifter could change its phase rapidly; handle high power; require control signals of little
power; be of low loss, light weight, small size, and reasonable cost; and have a long life. But in practice, no
one device is universal enough to meet all the requirements. Various phase shifters possess these properties
in varying degrees. Presently, three types of phase shifters are used in the phased-array antennas: the diode
phasers which are all reciprocal, the nonreciprocal ferrite phasers, and the reciprocal ferrite phasers. Each
of the three types has its own advantage. The choice of phase shifters is highly dependent on the radar
requirements.
Array Elements and Matching. Almost any type of radiating antenna element can be considered for an
array antenna. In practice, the dipole, open-ended waveguide (or small horn) and slotted waveguide have found
wide applications. Detailed descriptions of these antennas have been investigated in the standard textbooks
(1,5). However, note that the properties of a radiating element in an array is significantly different from its
properties when in free space. For example, the radiation resistance of a half-wavelength dipole in free space
is 73
, but when it is in an infinite array with half-wavelength spacing and a back screen of quarter-wave
separation, it will be 153
when the beam is broadside. The impedance also varies with scan angles.
The change of impedance with scan angle makes the matching of an array antenna difficult. There are
conditions where an antenna that is well matched at broadside may have some angles at which most of the
power is reflected. Unlike a conventional antenna, the mismatch of array antennas will affect both the level of
the radiated power and the shape of the antenna pattern.
Mutual Coupling. If two radiating elements are widely separated, the energy coupled between them is
small, and the influence of one to the other on the current excitation and pattern can be negligible. However,
when the elements are placed closer, their coupling will increase. Generally, the strength of the coupling in a
phased array is related to the distance between elements, the pattern of the elements, and the structure of the
array. For example, the radiation pattern of a dipole has a null at = 90 and a peak value at = 0 . Thus, the
dipoles in a straight line are loosely coupled, whereas the dipoles parallel to each other have a strong coupling.
For a phased array, the effect of coupling between elements in mainly focused on the pattern and impedance
of the element, which are usually called the active element pattern and the active element impedance. The
exact analysis of coupling in practical arrays is rather complicated. A convenient way is to assume that the
20
RADAR ANTENNAS
array is infinite in extent and has a uniform amplitude distribution and a linear phase taper from element
to element. Under such an assumption, every element in the array has the same environment; thereby a
significant simplification in the calculations can be made. It has been shown that the infinite-array model has
a good prediction to the actual arrays. Even for a modest array that has fewer than 100 elements, the predicted
results can also give reasonable agreement (23).
Errors in Arrays. As indicated in the section entitled Low-Sidelobe Arrays, the errors in the amplitude
and phase of the current at individual elements of the array may cause distortion of the radiation pattern.
Additional factors of the error include the missing or inoperative elements, rotation or translation of an element
from its correct position, and variations in the individual element patterns. These errors can result in a decrease
in gain, increase in sidelobes, and shift in the direction of the main beam.
When errors occur in the phase and amplitude of the aperture, the energy will be removed from the main
beam and distributed to the sidelobes. If the errors are purely random, they will produce random effects on the
main beam and all sidelobes. When the errors are correlated, the sidelobe energy will be lumped at discrete
locations in the far field. Usually, the correlated errors provide higher sidelobes than the random errors, but
they are located at some certain directions only. Both of the correlated and random errors should be considered
in the practical design of phased arrays (18,24).
Applications. In many cases, the phased-array antennas have been of considerable interest in the radar
systems because they have different properties from those of reflector antennas. Hence, a number of phasedarray radar systems have been built for different use (18,25). Examples include the AN/SPS-33 radar for the
purpose as aircraft surveillance from on-board ship; the AN/FPS-85 for satellite surveillance; the PAR and
MSR for ballistic missile defense; the AN/SPY-1 and PATRIOT for air defense; the EAR (Electronically Agile
Radar) for airborne bomber; the AN/TPN-19 and AN/TPS-32 used in aircraft landing systems; the AWACS and
AN/TPS-70 for airborne warning and control systems; the MESAR as a multifunction electronically scanned
adaptive radar; the AN/TPQ-37 Firefinder radar; the PAVE PAWS for providing early warning of ballistic
missiles and performing satellite tracking; the COBRA DANE for tracking of ballistic missile; and the COBRA
JUDY for collecting data on foreign ballistic missile tests (18,25).
Although the array antennas have many unique characteristics that make them candidates for use in
radar systems, they are expensive. As technology advances, the costs can be reduced, particularly in the areas
of phase shifters and drivers. In the meantime, the high demand for better performance with lower sidelobes
and wider bandwidth makes the costs high. In the future, the great potential for cost reduction may be the
application of solid-state systems with a transmit/receive module at each element.
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
R. C. Johnson, H. Jasik (eds.), Antennas Engineering Handbook, 2nd ed., New York: McGraw-Hill, 1984.
M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York: McGraw-Hill, 1990.
L. J. Cantafio (ed.), Space-Based Radar Handbook, Norwood, MA: Artech House, 1989.
M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York: McGraw-Hill, 1980.
C. A. Balanis, Antenna Theory Analysis and Design, 2nd ed., New York: Wiley, 1997.
M. J. Komen, Use simple equations to calculate beam width, Microwaves, 20 (13): 6163, 1981.
W. L. Stutzman, G. A. Thiele, Antenna Theory and Design, New York: Wiley, 1981.
R. L. Fante et al., A parabolic cylinder antenna with very low sidelobes, IEEE Trans. Antennas Propag., AP-28: 5359,
1980.
9. K. Aoki et al., Design method for an offset dual-shaped reflector antenna with high efficiency and an elliptical beam,
IEEE Proc. H. Microw. Antennas Propag., 140 (2): 121128, 1993.
10. Y. T. Lo, On the beam deviation factor of a parabolic reflector, IRE Trans. Antennas Propag., AP-8: 347349, 1960.
11. P. W. Hannan, P. A. Loth, A monopulse antenna having independent optimization of the sum and difference modes,
IRE Int. Conv. Rev., Part 1, 1961, pp. 5760.
RADAR ANTENNAS
21
12. C. Granet, Designing axially symmetric Cassegrain or Gregorian dual-reflector antennas from combinations of prescribed geometric parameters, IEEE Antennas Propag. Mag., 40 (2): 7682, 1998.
13. A. W. Love, Electromagnetic Horn Antenna, New York: IEEE Press, 1986.
14. H. E. Schrank, G. E. Evans, D. Davis, Reflector antennas, in M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York:
McGraw-Hill, 1990.
15. C. L. Gray, Estimating the effect of feed support member blocking on aperture gain and sidelobe level, Microwave J.,
(3), 8891, 1964.
16. J. Ruze, Antenna tolerance theoryA review, Proc. IEEE, 54: 633640, 1966.
17. J. D. Walton, Jr. (ed.), Radome Engineering Handbook, New York: Dekker, 1970.
18. T. C. Cheston, J. Frank, Phased array radar antennas, in M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York:
McGraw-Hill, 1990.
19. R. E. Willey, Space tapering of linear and planar arrays, IRE Trans. Antennas Propag., AP-10: 369377, 1962.
20. N. Goto, D. K. Cheng, Phase shifter thinning and sidelobe reduction for large phased array, IEEE Trans. Antennas
Propag., AP-24: 139143, 1976.
21. G. E. Evans, H. E. Schrank, Low sidelobe radar antennas, Microwave J., 26 (7): 109117, 1983.
22. M. R. Iravani, D. Maratukulam, Review of semiconductor-controlled (static) phase shifters for power systems applications, IEEE Trans. Power Syst., 9: 18331839, 1994.
23. J. Frank, Phased array antenna development, Appl. Phys. Lab. Rep., Johns Hopkins Univ., TG 882, March 1967.
24. J. L. Allen, The theory of array antennas, MIT Lincoln Lab. Rep. 323, July 1963.
25. P. E. Rawlinson, H. R. Ward, Radar antennas, in R. C. Johnson and H. Jasik (eds.), Antenna Engineering Handbook,
2nd ed., New York: McGraw-Hill, 1984.
RADAR CROSS-SECTION
Radar cross-section (RCS) is a measure of the magnitude of the scattered electromagnetic wave from a body
relative to the magnitude of the wave incident on the body. To many people RCS is synonymous with stealth.
Stealth technology has become an integral part of all military systems since it was first employed in the 1970s.
However, RCS is just one aspect of stealth, the design philosophy that seeks to make a platform (such as an
aircraft, ship, or ground vehicle) undetectable to a wide range of sensors. Such a platform is referred to as
low-observable (LO). In addition to radar, the infrared, acoustic, and visible signatures must be controlled.
In the past, the RCS has received the most attention because the detection range of radar is generally much
greater than that of other sensors. Undetectability does not necessarily imply a very low signature, only that
the target must be indiscernible against the background. In the visible region (wavelengths visible to the eye)
stealth has been used since World War II in the form of camouflage.
A typical radar and target arrangement is shown in Fig. 1. The transmitter illuminates the target from
a distance Rt . If the body is sufficiently far from the transmitting antenna, the incident spherical wave front
will be approximately planar across the target. The incident wave induces currents on the surface of the body,
and even within its volume if the material is not a perfect electric conductor. The induced currents give rise to
a scattered field, which is detected by the receiver at a distance Rr .
The arrangement shown in Fig. 1 is referred to as bistatic; the directions of the transmitter and receiver
are different as viewed from the target. The monostatic case is more common; that is, the directions of the
transmitter and receiver are the same as viewed from the target. For a truly monostatic radar the transmit
and receive channels should use the same antenna. Frequently the radar will use separate antennas for
transmit and receive, but they will be closely spaced on the same platform. This configuration is referred to as
quasimonostatic, and can be considered as monostatic for the purpose of specifying the RCS of a distant target.
The majority of radars currently in use for both military and civilian applications are monostatic.
Most radars are designed to provide information on the targets location and velocity. Location is given in
terms of the range (radial distance) from the radar along with direction. The range to the target is obtained by a
time delay measurement. An electromagnetic wave is transmitted in the direction of the target, and the round
trip travel time converted to distance using the known velocity of propagation of the wave. The direction of the
wave is obtained from antenna pointing. Velocity can be measured using the frequency shift due to the Doppler
effect or two range measurements at closely spaced times to obtain the targets range rate. Any measurement
of the target parameters assumes that the reflected wave is strong enough to be detected and processed by the
radar. Search radars are specifically designed for this purpose. They are used to scan large volumes of space
for targets at great distances (hundreds or even thousands of kilometers). Once a target has been detected it is
handed over to a tracking radar. Modern multifunction radars are designed to perform both search and track
functions.
Although both the search and track operations are more difficult for targets with low RCS, the primary
objective of RCS reduction is to defeat search radars (i.e., avoid detection). The role of the RCS in the detection
of a target is demonstrated by the basic form of the radar equation derived later. It shows that when all other
1
RADAR CROSS-SECTION
Fig. 1. Bistatic RCS refers to an arrangement where the transmitting and receiving antennas are at different locations
as viewed from the target. For the monostatic case the transmitting and receiving antennas are co-located.
radar system parameters are fixed, the maximum detection range of a target varies as the fourth root of RCS.
Thus by reducing the target RCS, the range at which it is first seen by the radar is also reduced.
The RCS of modern aircraft and ships has been significantly reduced, but in response to the lower cross
sections, more sensitive radars have evolved. The RCS designer is at a disadvantage in this respect because
a truly LO platform must be designed from the start. Stealth cannot simply be added on after all other
design tradeoffs have been made. The entire design and test cycle of an aircraft can be a decade, and its
operational life several decades. On the other hand, the radar designer may be able to significantly increase
the capability of a radar by simply replacing a system component such as an antenna or signal processor.
These are relatively minor changes and can be integrated into deployed radars in a couple of years, thereby
decreasing the effectiveness of the LO platform for the rest of its operational lifetime.
The scattering properties of the target are only one of several considerations that enter into the selection of
the operating frequency of a radar (1,2). Others are size constraints, antenna gain and beam width, transmitter
power, ambient noise, Doppler shift, and atmospheric attenuation. The frequency spectrum is divided into bands
with letter designations as shown in Table 1. A few long-range ballistic missile defense radars operate in the
300 MHz region (UHF), but most others use frequencies greater than 1 GHz (L band and above). Devices at
low frequencies are larger than similar devices at a higher frequency, and therefore the systems components
require more volume. Likewise the antenna must be larger at a low frequency than at a high frequency to
achieve the same beam width and gain. Low frequencies are capable of handling more power because the
applied voltages can be higher without causing breakdown. Finally, ambient noise is lowest in the 1 to 10 GHz
range, and low-altitude atmospheric attenuation favors frequencies below 18 GHz.
RADAR CROSS-SECTION
The function of the radar also influences the choice of frequency. For instance, an airborne imaging radar
must be compact and lightweight, but also have a narrow antenna beam for resolving scatters. These all favor
a high frequency. On the other hand, a ground-based search radar must radiate high power to achieve a large
detection range. In this case the larger and heavier equipment that is necessary for high-power operation is
not a problem.
Implicit in the definition of RCS is the assumption that the frequency is fixed.
Mathematically the RCS is given by
where
Ei = plane wave electric field intensity incident from the direction (i , i )
Es = scattered spherical wave electric field intensity observed from the direction (, )
Rr = distance from the target to the observer as shown in Fig. 1
RADAR CROSS-SECTION
At first glance is may appear that RCS is a function of range because of the R2 r in Eq. (1). However, the
scattered field is a spherical wave whose electric field has a 1/Rr dependence, which, when squared, cancels the
factor in the denominator. Hence RCS is a range-independent quantity.
Electromagnetic scattering and radiation problems can be formulated in either the time domain (TD) or
the frequency domain (FD) (4,5). In fact, the scattering body can be represented as a linear system with impulse
and frequency responses. The RCS as a function of time is related to the RCS as a function of frequency by
the Fourier transformation. There are advantages and disadvantages to working in each domain. Early radar
systems operated over narrow frequency ranges, and the need to compare calculated and measured data
dictated that calculations also be done in the frequency domain. Modern radars operate over wide frequency
ranges, and in order to characterize their performance adequately, the target RCS must also be determined
over the same wide range of frequencies. For a large, complex target, the calculation of the RCS at each
frequency is potentially a time-consuming task. It may be more efficient to compute the targets response to a
waveform in the time domain and then perform a Fourier transformation to obtain the RCS at many frequencies
simultaneously. Thus the measurement or calculation of RCS can be performed by either of two procedures:
(1) Frequency Domain The received power is obtained as the target is rotated through a range of angles while
a continuous wave (CW) fixed-frequency source is radiating. The frequency can be changed and another plot
of RCS versus angle obtained, and so on. At any fixed angle the RCS from a large number of frequencies
can be Fourier transformed to obtain the time-domain response of the target at that particular angle.
(2) Time Domain The received power as a function of time is obtained at each aspect angle of interest. The
time histories are inverse Fourier transformed to yield the frequency response of the target at each angle.
If the Fourier transform of the waveform and the impulse response of the radar system are known, then
the targets frequency response can be converted to RCS.
There are several aspects of the RCS that enter into the design of an LO platform, and therefore the
discussion of the RCS logically falls along similar lines. They include the topics:
(1)
(2)
(3)
(4)
A brief overview of each of these four topics is given, followed by mathematically based discussions of RCS
prediction and reduction.
RADAR CROSS-SECTION
RADAR CROSS-SECTION
Fig. 2. The RCS of a sphere illustrates the three frequency regions: (1) Rayleigh (linear region at low frequencies), (2)
resonance (oscillatory region), and (3) optical (flat high frequency region).
polarizations are identical. However, measuring both components provides the most information available
from the scattered field and may assist in target identification.
A polarization-specific RCS can be expressed as
where p denotes the polarization of the incident wave and q the polarization of the receiver. Common polarization designations are defined with regard to a reference such as the ground or surface. They include: horizontal
and vertical (H and V); parallel and perpendicular ( and ); rectangular (x and y). A scattering matrix relates
the scattered field components to the incident field components:
RADAR CROSS-SECTION
Fig. 3. A cylinder illustrates how a targets RCS can depend on the incident wave polarization. At low frequencies the
cylinder becomes a thin wire that is transparent to a perpendicularly polarized wave.
Fig. 4. Simple geometric shapes such as plates, cylinders, and spheres can be used to build more complicated targets. The
interaction between the individual parts is important in determining low-level RCS.
Scattering Mechanisms
The RCSs of targets encountered by most radars are complicated geometrical shapes. There are a few exceptions
such as weather balloons and buoys. However, simple shapes such as plates, spheres, cylinders and wires are
useful in studying the phenomenology of RCSs. Furthermore, complex shapes can be decomposed into basic
geometrical building blocks (primitives) that can be assembled to form a more complex shape. Figure 4 shows
an aircraft represented by cylinders, plates, cones, and hemispheres. The RCS of the model in Fig. 4 is shown in
Fig. 5. A collection of basic shapes gives an acceptable RCS estimate that can be used during the initial design
stages of a platform. The locations and levels of the largest RCS lobes are of most concern at this stage of the
design process. The accuracy of the RCS estimate at other angles depends on how the interactions between
the various shapes are handled. These are more difficult to include, as will be seen later. Even when they
are included, the agreement with the measured RCS may not be particularly good if the interactions are not
modeled accurately. For an LO target more sophisticated numerical techniques must be used.
As evident from the pattern in Fig. 5, the RCS of a large, complex target is a rapidly varying function of
angle. However, as complicated as it looks, isolated features of the pattern shape and level can be associated
with the behavior of the currents induced in and on the scattering body. For example, the high narrow spikes
RADAR CROSS-SECTION
Fig. 5. An azimuth plot of the RCS of the aircraft in Figure 4 is shown. The two large lobes are due to the fuselage, which
is modeled as a cylinder.
Fig. 6. The total RCS of a target can often be decomposed into a sum of contributions from distinct scattering mechanisms.
Some important mechanisms for electrically large targets are depicted in this figure.
result from mirrorlike specular reflections from large flat surfaces. Other lobes can be associated with surface
waves, multiple reflections, diffraction, and so on. These so-called scattering mechanisms are depicted in Fig.
6. The lines in the figure are ray representations for propagating waves. They denote the direction of power
flow and are normal to the planes of constant phase.
For any particular aspect angle, frequency, and polarization, if one scattering mechanism dominates over
all others, then it may be possible to express the total scattered field as a combination of scattered fields Es1 ,
Es2 , Es3 , . . ., each due to a separate scattering mechanism:
RADAR CROSS-SECTION
It follows from Eq. (1) that the total RCS can be written as a sum of RCSs, each attributed to a separate
scattering mechanism:
Note that Eq. (5) is a convenient way of decomposing the scattered field for the purpose of assisting the
RCS modeling and reduction efforts. First, if sufficiently accurate models exist for the individual scattering
mechanisms, then Eq. (3) suggests that they can be combined under the appropriate circumstances to model
their collective effects for a complex target. Secondly, since the largest terms in Eq. (3) correspond to the source
of the most intense scattering, it may be possible to determine which scattering mechanism dominates, and
thereby choose an appropriate reduction technique.
Reection. This is the mechanism that yields the highest RCS peaks, but they are limited in number
because Snells law must be satisfied. When multiple surfaces are present, multiple reflections are possible.
For instance, the incident plane wave could reflect off of the fuselage of an aircraft, hit a fin, and then return
to the radar.
Diffraction. Diffracted fields are those scattered from discontinuities such as edges, corners, and vertices. The waves that are diffracted from these shapes are less intense than reflected waves, but they can
emerge over a wide range of angles. At aspects requiring low RCS, the diffracted waves can be significant.
Surface Waves. The term surface wave refers to the current traveling along a body and includes several
types of waves (7). In general, the target acts as a transmission line, guiding the wave along its surface. If
the surface is a smooth closed shape such as a sphere, the wave can circulate around the body many times.
Such waves are called creeping waves because they appear to creep around the back of a curved body. On
curved bodies the surface wave continuously radiates. Radiating surface waves on flat bodies are usually called
leaky waves. Traveling waves appear on slender bodies and along edges, and suffer little attenuation as they
propagate (8). If the surface is terminated with a discontinuity such as an edge, the traveling wave will be
reflected back along the surface and radiate towards its origin.
Ducting. Also called waveguide modes, ducting occurs when a wave is trapped in a partially closed
structure. An example is an air inlet cavity on a jet. Once the wave enters the cavity, many bounces can occur
before a ray emerges. There are many paths that the ray can take, and therefore rays emerge at almost all
angles. The result is a large, broad monostatic RCS lobe. An optical analogy of this is the glowing of a cats eye
when it is illuminated by a light.
Interactions. It is possible that these mechanisms will interact with each other. For example, a wave
reflected from a flat surface can subsequently be diffracted from an edge or enter a cavity. For a complex target
the interactions are not always obvious and the decomposition of the scattered field no longer makes sense.
Prediction Methods
The analytical methods used to calculate RCSs are similar to those used in antenna analysis. A radar target is
essentially acting as an antenna. The incident wave induces a current on the target, and the induced current
radiates a field just as an antenna would. However, in the radar case it is called a scattered field as opposed to
a radiated field. A good example is a reflector antenna or satellite dish. The incident field from a feed horn
excites currents on the reflector, which in turn give rise to a radiated field. In the radar analogy, the feed horn
represents the radars transmit antenna, while the reflector is the target. If the distribution of current is known
for a target, then the currents can be used in the radiation integrals (StrattonChu integrals) to determine the
scattered field. Once the scattered field is known, the RCS can be computed.
10
RADAR CROSS-SECTION
Of course, the problem is that the induced current on the target is unknown and generally very difficult
to determine. An analytical solution is possible in only a few simple cases. In other cases it is feasible to solve
Maxwells equations or integral equations for the current numerically. These solutions are rigorous in that there
are no assumptions or restrictions imposed. The only error in the solution is due to the numerical evaluation
of the integral or differential equations. The alternative to a numerical solution is to take an intuitive guess at
the current, in which case the resulting RCS is only approximate.
Several methods of RCS prediction are described below. The classical solution techniques are not discussed
in this list, because most are limited to one- or two-dimensional structures or simple three dimensional ones.
The methods of interest here are those that can be applied to complex three-dimensional targets. The methods
most commonly encountered are physical optics, microwave optics (ray tracing), the method of moments, and
finite-difference methods.
Physical Optics. The physical optics (PO) approximation (9) provides a means of estimating the surface
current induced on an arbitrary body. On the portions of the body that are directly illuminated by the incident
field, the induced current is simply proportional to the incident magnetic field intensity. On the shadowed
portion of the target the current is assumed to be zero. The current is then used in the radiation integrals to
compute the scattered field far from the target.
PO is a high-frequency approximation that gives best results for electrically large bodies (L 10). It is
most accurate in the specular direction. Since PO assumes that the current abruptly decays to zero at a shadow
boundary, the computed field values at angles far from the specular directions and in the shadow regions are
inaccurate. Furthermore, surface waves are not included. PO can be used in either the time or the frequency
domain. It can be supplemented by the physical theory of diffraction (PTD), which provides corrections to PO
approximation that improve the accuracy of the current distribution near edges.
Microwave Optics. Microwave optics refers to a collection of ray-tracing methods that can be used
to analyze electrically large targets of arbitrary shape. The rules for ray tracing in a simple medium (linear,
homogeneous, and isotropic) are similar to reflection and refraction in optics. Geometrical optics (GO) (10) is
the classical theory of ray tracing used since the days of Newton. It provides a formula for computing the
reflected and refracted fields. In addition, the geometrical theory of diffraction (GTD) is used in conjunction
with GO to include diffraction. Diffracted rays originate from the scattering of the incident wave at edges,
corners, and vertices. The formulas are derived on the basis of infinite frequency ( ), which implies an
electrically large target. Ray optics is frequently used in situations that severely violate this restriction and
still yields surprisingly good results. The major disadvantage of ray tracing is the bookkeeping required for a
complex target. It is used primarily in the frequency domain.
Method of Moments. The method of moments (MM) is a technique used to solve an integral equation
(11). Integral equations are so named because the unknown quantity appears in an integrand. In electromagnetics they are derived from Maxwells equations and the boundary conditions. The unknown quantity can be a
combination of volume or surface current (either electric or magnetic). The MM reduces the integral equations
to a set of simultaneous linear equations that can be solved using standard matrix algebra. The size of the
matrix involved depends on the size of the body; current computer capabilities allow bodies on the order of 10
or 20 wavelengths to be modeled.
Most MM formulations require a discretization (segmentation) of the body. Therefore, they are compatible
with finite-element methods used in structural engineering, and the two are frequently used in tandem during
the design of a platform. The MM can be used to solve both time- and frequency-domain integral equations.
Finite-Difference Methods. Finite differences are used to approximate the differential operators in
Maxwells equations in either the time or the frequency domain (5,6). As with the MM technique, the target
and surrounding space must be discretized. Maxwells equations and the boundary conditions are enforced on
the surface of the target and at the boundaries of the discretization cells. This method has found extensive use
in computing the transient response of targets to various waveforms. Finite-difference methods do not require
the large matrices that the MM does, because the solution is stepped in time throughout the scattering body.
RADAR CROSS-SECTION
11
where A is the area of the surface. Thus the RCS can be minimized by minimizing the area presented to the
radar. If it is not possible to tilt a large surface sufficiently, then it can be broken into a number of smaller
surfaces, each tilted in a slightly different direction.
These approaches have led to rather unconventional aircraft shapes such as that of the F-117 Nighthawk
(commonly referred to as the Stealth Fighter) (12). To date, shaping has not been applied extensively to ships
and ground vehicles, although that is beginning to be done. Modern ship designs are making some concessions
to RCS control, such as canting deckhouse walls and enclosing mast structures.
Another aspect of shaping is edge alignment. Once the specular reflections have been controlled, the
contributions from edges (such as traveling waves and diffraction) may become important. Aligning edges is
one way of maintaining low-RCS spatial sectors. For example, in plan form it is seen that the edges of the B2
bomber fall along only a couple of fixed angles.
The second rule of shaping is to avoid or hide retroreflectors. Retroreflectors are simple geometrical
structures that tend to redirect the incident radar wave back in the direction from which it arrives. Surfaces
that meet at a 90 angle are strong retroreflectors, and therefore the angles between the tail rudder and
elevator surfaces of a stealthy aircraft are always canted. New designs such as the tailless McDonnell Douglas
X-36 have no vertical surfaces at all.
Cavities are another type of retroreflector. They have a very broad scattering pattern. Thus jet intakes are
often located above the wing or else employ an electromagnetic screen to prevent penetration by the incident
wave.
Obviously there is a limit to the degree of shaping that can be incorporated while simultaneously meeting
other operational requirements. For example, extreme shaping applied to an aircraft degrades its aerodynamic
performance and control characteristics. Another disadvantage of low-RCS shapes is that they have slim
profiles, which reduces the useful enclosed volume relative to a conventional shape.
Materials and Coatings. Obviously the material composition of a body affects its RCS. For example,
a metal plate has a higher RCS than a glass plate. Similarly, a plastic aircraft body has a lower RCS than one
with a metal skin. The electrical characteristics of a material are defined by its constitutive parameters (13):
(1) Permittivity, = 0 r = j (F/m)
(2) Permeability, = 0 r = j (H/m)
(3) Conductivity, c (/m)
12
RADAR CROSS-SECTION
where 0 and 0 are the permittivity and permeability of free space. The imaginary parts are associated
with dissipation (attenuation) of the fields in the medium.
The permittivity determines the fundamental relationship between the electric field intensity E and flux
density D. For the vast majority of materials that occur in nature one can write
where the boldface denotes a matrix. Expanding Eq. (7) and explicitly including the spatial and frequency
dependences gives
Likewise, the permeability determines the fundamental relationship between the magnetic field intensity H
and magnetic flux density B, which for most materials can be written as
or
RADAR CROSS-SECTION
13
and
From Eq. 1 the RCS is proportional to the square of the scattered electric field intensity. Since the scattered
field intensity (in this case the reflected field) is proportional to , it follows that | |2 . Thus the RCS will
be reduced in direct proportion to the reduction in reflection coefficient squared. This is an ideal limit, because
in practice the material thickness is finite, and there will likely be some reflection from its back face and
scattering from its edges.
Unfortunately, in most applications the target materials cannot be selected solely on the basis of the
electromagnetic properties r and r . Mechanical properties such as strength, weight, and thermal conductivity
have a higher priority. In the drive to reduce weight and increase strength, composite materials have found
widespread military and civilian application (14).
Composites are formed from multiple constitutive materials in a manner that yields a final product having
desirable properties from each material. Graphite is an example of a composite that is used extensively in the
aircraft industry. Although the primary motivation for the use of composites has been structural, the reflection
from most composite materials is lower than that for conductors. However, the majority of the incident field
that is not reflected from the composites surface is transmitted through the material, and the transmitted field
can potentially be scattered from internal metallic objects. The use of nonconductors also complicates other
electrical design issues such as the grounding of electrical systems and lightning protection.
An absorbing layer is another example of the application of materials for RCS reduction. A layer of
absorbing material is deposited on the surface of the target. As the incident wave propagates through the
material, energy is extracted from the field and its intensity diminishes (15). Thus a weaker field is reflected
at the target surface, and the reflected wave is attenuated further as it travels back through the material. The
circuit analog of this type of material is a resistor.
A material that performs the function just described is referred to as a radar absorbing material (RAM).
High absorption is desired, yet the material thickness and weight must be minimized. Unfortunately, many
materials with high loss also have high reflection coefficients. Thus there is significant reflection of the incident
wave at the absorbing-layer face, and little energy enters the material where attenuation takes place.
There are some materials that possess more complicated or exotic electromagnetic properties that can
potentially be exploited for RCS reduction. For example, since the pq affect the velocity of propagation along
the coordinate axes of the material, it may be possible to construct for a material so that the polarization
of the scattered field is different than that of the incident field. Thus the polarization of a wave rotates as it
propagates through the medium, and scattered energy is converted to a field component that the radar is not
capable of sensing.
14
RADAR CROSS-SECTION
Another example is a spatially dispersive medium. The electric flux density depends on both the electric
field intensity and the magnetic flux density. A chiral material (16) has the constitutive relationship
This type of behavior is associated with materials that have a helical molecular structure. It is possible to
produce this type of behavior artificially by dispersing small helical coils throughout a bulk material.
In the past, the reduction of RCSs has focused on treating each scattering source, or hot spot, individually.
For instance, an absorbing layer might be installed to reduce scattering from a wing edge. The layer will reduce
the edge scattering, but will also introduce new scattering sources at the perimeter. For a useful reduction
method, the new scattering sources must be much weaker than those of the original edge. But as the scattering
intensities of the sources become smaller, one quickly reaches the point where the new scattering sources
introduced by the treatments are comparable to those being treated. At this level the interaction between
scatterers becomes important, and this complicates the design and application of RCS treatments.
Cancellation Techniques. Cancellation refers to RCS reduction methods that are based on the introduction of a secondary scatterer whose scattered field is adjusted to cancel the reflection of the primary target
(17,18). The total field at the radar receiver is the vector sum of the fields scattered by the target and the
cancellation element. Therefore, if the cancellation field can be made equal and opposite to the field scattered
from the body at the angle, polarization, and frequency of interest, then destructive interference occurs. The
secondary scatterer is referred to as the cancellation element, and its scattered field the cancellation field.
The cancellation element can be a detached body located near the original target, or an attached protrusion. Both are referred to as parasitic elements. The combination of the target with the scattering element is
frequently called the loaded body. If the cancellation field is fixed, then the system is a passive one. On the
other hand, if the system can respond to changes in the incident field, then it is an active one. A fair amount
of information about the radar and the target is required for an active system: the threat radar frequency, its
direction and polarization, and the targets RCS in the direction of the radar.
Passive Cancellation. Since the cancellation of the two fields is essentially a tuning method, this technique is only effective over a narrow frequency band and usually limited to a small spatial sector. If large
parasitic elements are to be avoided, then the magnitude of RCS that can be canceled is small. Thus passive
cancellation is usually used to supplement shaping and absorbers.
The most common cancellation elements are slots and wires. Two orthogonal elements are required to
cancel an incident wave of arbitrary polarization. The elements can be increased in size or number so as to
provide sufficient gain to cancel the RCS in the direction of interest.
Two examples of passive cancellation are the Salisbury screen and Dallenbach layer (19). A Salisbury
screen consists of a resistive film located a distance d in front of the target surface. The film resistivity is chosen
so that the reflection from its surface is exactly canceled by the field that passes through the film, is reflected
from the target, and then is transmitted back through the film.
The Dallenbach layer shown in Fig. 7 is similar to the RAM coating discussed earlier, except that the
layer thickness is chosen to provide destructive interference between the fields scattered from the top of the
layer and the boundary at the back of the layer, rather than absorption within it. Thus the top of the layer
serves the same purpose as the resistive film does for the Salisbury screen.
Active Cancellation. Active cancellation (also known as adaptive cancellation) is the extension of passive
cancellation to handle dynamic threat scenarios. In the context used here, it does not include deception, which
comprises techniques of modifying and retransmitting a signal. For the present discussion, the source of RF
energy is the radars signal. Two levels of sophistication are considered:
RADAR CROSS-SECTION
15
Fig. 7. A Dallenbach layer is used to reduce the RCS of a conductor. The material electrical characteristics and thickness
are chosen to provide a reflection coefficient of zero.
Fig. 8. Traveling waves along a surface are treated using a transmission line matching method: the quarter wave
transformer.
(1) Fully active The cancellation network receives, amplifies, and retransmits the threat signal in such a way
that it is out of phase with the static (skin) RCS of the target. The transmitted signal amplitude, phase,
frequency, and polarization can be adjusted to compensate for changing threat parameters.
(2) Semiactive No amplification of the incident signal energy is provided by the cancellation network, but
adjustable devices in the network allow the reradiated signal to compensate for limited changes in the
radar signal parameters.
The demands for a fully active system are almost always so severe as to make it difficult to implement.
It requires amplifiers and antennas that cover the anticipated threat angles, frequencies, incident power
densities, and polarizations. A knowledge of the radars direction is required, as well as the targets own RCS.
A semiactive system is not as complicated with regard to hardware, but the use of adjustable devices still
requires bias lines, controller units, and a computer with the appropriate databases.
Because of the growing importance of low-level scattering sources (in particular, manufacturing-induced
errors that differ from platform to platform), adaptive cancellation techniques are receiving increased attention. Adaptive cancelers have the ability to modify the amplitude and phase of the new scattering source to
compensate for changes in the original scattering source that may occur with frequency or observation angle. In
principle, adaptive techniques are capable of adjusting to changing physical and electrical conditions brought
on by exposure to the environment. However, cancellation is most effective when there are a small number of
coherent scattering sources rather than a large number of random sources.
Treatments for Surface Waves. Surface waves, and in particular traveling waves, present a special
challenge as far as RCS reduction is concerned (6). The target acts as a combination of antenna and transmission
line, collecting incident energy and guiding it along the surface. As depicted in Fig. 8, when a discontinuity
(such as a termination) is encountered, the surface wave is reflected. As it propagates in the reverse direction,
it loses energy by radiation. A dielectric-coated ground plane can support a surface wave for incident plane
waves having a component of E normal to the surface; a magnetic-coated ground plane can support a surface
wave for incident plane waves having a component of H normal to the surface.
The most common technique for treating surface-wave reflections that arise from surface discontinuities
is one borrowed from transmission line theory: the use of transformer sections. Typically a quarter-wave
matching section is appended to the termination of the surface, and the surface characteristics of the matching
16
RADAR CROSS-SECTION
section are chosen to allow the wave to continue propagating in the forward direction, thereby eliminating the
reflected field.
RADAR CROSS-SECTION
17
improved by adding higher-order corrections based on ever more detailed electromagnetic models as well as
empirical data. Both the rigorous and the approximate computer models have incorporated geometry builders
that can utilize computer-aided drawing (CAD) files generated by commercially available drawing programs.
This capability allows for concurrent engineeringthe ability to make a structural change and immediately
assess its effect on the RCS.
Effect of Imperfections and Flaws. An important limitation in achieving low RCS is the imperfections
introduced during the manufacture and assembly of the platform and the installation of its systems. Their
effect on RCS performance depends on whether the imperfections are periodic or random in nature. Most error
sources are nearly random, and their net effect on RCSs is to shift energy from the scattering mechanisms
discussed previously to diffuse scattering. Diffuse surfaces reflect uniformly in all directions, independent of the
angle of incidence; they are electrically rough surfaces. The magnitude of the diffuse scattering increases with
the extent of the surface roughness (imperfection). Since diffuse scattering is not angle-dependent, shaping is
not a practical way of reducing the RCS for it, and therefore very close tolerances must be maintained to keep
the diffuse scattering to acceptable levels. Maintaining close tolerances is one of the primary cost drivers in
the manufacture of LO platforms. Eventually a point is reached where the question must be asked: How much
money is an additional RCS reduction of a few percent worth?
Maintenance and Testing. A second important cost driver is the maintenance of an LO platform.
The natural elements, along with the harsh operating environment, stress and degrade the electrical and
mechanical properties of materials. Special testing methods are required to determine if and when the RCS
has been affected by exposure to the environment. So-called gono-go tests compare a set of measured data from
the platform under test with a reference data set obtained from a known clean one of identical configuration.
A significant deviation in the data sets would signify a correspondingly significant change in RCS. A test of this
type does not require a high level of precision, and therefore can be conducted using portable test equipment.
Data can be taken by probes located close to the platform surfaces, that is, in the near-field region of the
scatterers.
If a platform fails the gono-go check in the field, then it will be sent to a test and repair facility
for further evaluation. Specialized facilities and repair techniques are required because of the uncommon
materials involved and the exacting tolerances that must be satisfied. The integrity of localized repairs can be
checked using hand-held reflection devices that verify local surface reflection characteristics.
Tradeoff between Stealth and Electronic Warfare. Low observability demands the most of technology and forces concessions in many other aspects of platform design and performance. At some point the
burden of increasing stealth may become unacceptable. Electronic warfare (EW) techniques (now referred to
as electronic attack and electronic protect in the United States) are a complement to low RCS. Common EW
techniques currently in use around the world include jamming, decoys, and chaff (22). The major disadvantage
of EW is that the radar is aware of the targets presence. However, the radar is denied information on the
targets range and velocity. Even though the radar is not completely disabled, a combination of EW and low
RCS can cause the radar to loose lock, forcing it to reacquire a target repeatedly. This is a time-consuming
process that ties up valuable radar resources. With all other factors constant, chaff, deception, and jamming
are more effective for low-RCS targets than for conventional ones.
Some older platforms currently in the field are being retro-fitted or upgraded to reduce the RCS and other
signatures. All new aircraft, ships, and ground vehicles are incorporating some degree of stealth design and
technology. However, it appears that the complete dominance of stealth in all aspects of design, as seen in the
F-117, has been abandoned. Programs like the F-22, the Comanche helicopter, and the Euro-fighter are opting
for a balance between stealth and other performance measures.
18
RADAR CROSS-SECTION
The power scattered by the target that is delivered to the receive antenna terminals is given by
The first factor on the right-hand side is the power density incident on the target. Multiplying this factor by
gives the fraction of incident power scattered in the direction of the receiver. The product of the first three
factors represents the scattered power density at the receive antenna. Finally, multiplying by Aer gives the
scattered power collected by the antenna.
Specializing to the monostatic case (Gt = Gr G and (Rt = Rr R) and using the relationship between
effective area and gain give the result
This form of the radar equation is too simple to be of practical use in predicting a radars performance. However,
two important features are evident. The first is the R4 in the denominator, which is a disadvantage to the radar
system designer, whose goal is to increase the detection range of the radar. The second is that the received
power is linearly related to , which is a disadvantage to the RCS engineer.
If the minimum scattered power that the radar receiver can detect is Pm , then the corresponding maximum
detection range is obtained by solving the radar equation for R:
Therefore, to halve the maximum detection range requires that the RCS be reduced by a factor of 24 (=16).
Finally, note that the radar equation was derived for a single-frequency wave. Typical radar waveforms,
such as pulse trains, occupy a band of frequencies. The radar equation may have to be evaluated at several
frequencies in the radar band if the quantities in the equation are not constant.
RADAR CROSS-SECTION
19
All quantities appearing throughout this article are phasors unless otherwise indicated.
Spherical and Plane Waves. Ideal spherical waves (i.e., waves for which the planes of constant phase
are spheres) originate from point sources, which do not actually exist. However, for practical purposes, spherical
wave behavior results if the observer is located far enough from an extended source as depicted in Fig. 1.
Spherical waves have the phasor form
where r is the distance from the source. The vector constant E0 (which may have complex components) is
determined by the amplitude, phase, and polarization of the extended source.
At great distances from the source (r ) the target only intercepts a small section of the spherical wave
front with a very large radius of curvature. Thus it appears that planes of constant phase exist over the extent
of the target, and the electric field can be approximated by
where r is now the distance from the phase reference and E0 is determined from the intensity of the spherical
wave at the target. If the source is at the origin of a spherical coordinate system, then E0 can only have and
components in the far zone. (This would be the case for a radar antenna located at the origin.)
Both spherical and plane waves have the characteristics of a transverse electromagnetic (TEM) wave:
(1) The electric and magnetic field vectors are perpendicular to each other and the direction of propagation.
(2) The magnitudes of the electric and magnetic fields are related by the intrinsic impedance of free space.
20
RADAR CROSS-SECTION
Fig. 9. Current flows on the surface of a PEC. The total scattered field at the observation point is obtained by an integration
of the current over the entire surface.
The corresponding magnetic field intensity for a plane wave propagating inward toward the origin ( = r) is
The component of the electric field is called transverse magnetic to z (TMz ) because the magnetic field vector
associated with this component [E0 term in Eq. (16)] lies entirely in a plane transverse to the z axis. Similarly,
component of the electric field is referred to as transverse electric to z (TEz ) because this component of the
the
electric field [E0 term in Eq. (15)] lies entirely in a plane transverse to the z axis. Thus, for a TMz -polarized
plane wave E0 = 0; for a TEz -polarized plane wave E0 = 0. Any arbitrary polarization can be expressed as a
linear combination of the two. In a similar manner one could define TEx , TMy , and so on.
Radiation Integrals. Scattered fields are set up by a combination of electric and (in general) magnetic
currents. Currents are induced such that the total fields satisfy the boundary conditions and Maxwells equations. The radiation integrals provide a means of calculating the fields due to a prescribed set of currents.
The radiation integrals, or StrattonChu integrals, are integral solutions to Maxwells equations. They can be
derived directly by taking the curl of Maxwells first two equations, using the vector Greens theorem, and then
integrating (23).
For RCS calculations, according to Eq. (1), only the scattered field far from the target is of interest,
because R . Consider a scattering body of volume V that is enclosed by a surface S as shown in Fig.
9. The standard convention is to denote observation point quantities by unprimed symbols, and source point
quantities by primed symbols.
RADAR CROSS-SECTION
21
Fig. 10. If a target is not a PEC, then volume currents may exist inside of the target. The volume currents can be replaced
by surface currents using the surface equivalence principle.
where J (A/m2 ) and M (V/m2 ) are the volume electric and magnetic currents, respectively. (Magnetic current is
a fictitious quantity because isolated magnetic charge does not exist, but under some circumstances it is used
as a matter of mathematical convenience.)
The use of volume current is advantageous for computing the scattered fields from highly inhomogeneous
bodies. However, for most targets, it is more efficient to use a surface current formulation. As shown in Fig.
10, current flows on all interfaces between materials, as well as the external surface of the target. For surface
currents, the far-field form of the radiation integral becomes
where S includes all of the surfaces over which nonzero current exists. If the current is known, then the
integrals in Eqs. (17) or (18) can be evaluated numerically.
22
RADAR CROSS-SECTION
If the scattering body is a perfect electrical conductor (PEC), then only electric current exists, and the
surface of integration corresponds to a real physical surface. However, S in Eq. (18) need not exist; it can be
a fictitious surface defined for convenience. The currents on the fictitious surface can be determined from the
actual tangential electric and magnetic fields at every point on S. Thus
where n
is the unit vector normal to the surface. This is a consequence of the equivalence principle of electromagnetics (4).
Approximate Prediction Methods. Approximate prediction methods are used for a variety of reasons,
but primarily because they are computationally less demanding, thereby allowing data to be generated more
rapidly. This is an advantage in the early stages of design, when major tradeoff studies are being performed.
Coarse approximations can be applied in the beginning, and then refined as the design progresses. Eventually
it may be necessary to verify the performance of a specific design by a rigorous calculation.
At microwave frequencies, useful approximations should include scattering contributions from large surface areas, edges, surface waves, and interactions between and among surfaces and edges. There are two
approaches to computing the RCS. One is based on estimating the induced currents and is referred to as a
current-based approach: the PO approximation. The second is based on postulating relationships between the
incident and scattered waves, or rays: GO. This is a ray-based approach.
PO and GO are used to calculate the scattered field from surfaces; the effects of the edges of surfaces are
not properly modeled or not included. Supplemental theories have been developed to allow for edge scattering.
The PTD is the current-based method used in conjunction with PO. The GTD is a ray-based method for
calculating diffracted fields. Both the current and ray approaches can be shown to yield identical results under
the same conditions. Table 3 summarizes the two methods.
Physical Optics. A rigorous calculation of the currents J s and M s typically requires a great amount
computational effort. The PO approximation is an estimate of the electric current that is accurate for a large,
RADAR CROSS-SECTION
23
where
and
The subscript i denotes incident wave quantities. The x, y, and direction cosines are defined by
24
RADAR CROSS-SECTION
Fig. 11. A comparison of the RCS of a plate with dimensions a = 5 and b = 2.5 in the plane = 0 . The approximate
(PO) and rigorous (MM) methods are in close agreement in the region of the main lobe but disagree at wide angles.
For the monostatic RCS ui = u and vi = v; therefore the exponents in the integrand can be combined. Furthermore, the x and y integrals can be separated, and both are of the form
Now assume that a TMz -polarized plane wave is incident (E0 = 0) and the radar receive antenna is sensitive
to the same polarization. Thus we use the notation where the first subscript denotes the polarization of the
receiver and the second the polarization of the incident wave. Using Eqs. (23) and (24) in the definition of RCS
yields
RADAR CROSS-SECTION
25
then the edge-scattered field can be computed by using this current in the radiation integral. The total field at
an observation point is the sum of the PO and edge-scattered fields.
For a surface that is very large relative to the wavelength of the incident wave, the current distribution at
a point near the edge is primarily determined by the local features of the geometry. If the edge is very long and
the point under consideration is not close to the end of the edge, then the current is approximately the same as
that for an infinitely long edge. Thus edge scattering can be considered a local phenomenon, and the current
approximated by the current of the corresponding canonical scattering problem. (A canonical problem is one
that has fundamentally the same geometry as the one under consideration, but whose solution is known.) If
the total current can be determined for a canonical problem, in this case an infinite knife edge, then the PO
component can be subtracted out and the result used as an estimate for the fringe current on a finite edge. The
fringe currents are then used in the radiation integrals to obtain an edge-scattered field.
This approach has the advantage that the edge-scattered fields remain finite at shadow and reflection
boundaries. However, the fringe currents extend a significant distance from the edge (a minimum of about
0.5), so that two-dimensional integrals must be evaluated. Michaeli has reduced the PTD fringe currents to
filaments that flow along the body edge, thus reducing the integration to one dimension (25).
The electric and magnetic current filaments for a knife edge are given by
where
The PTD is simple in principle, but can be inconvenient in practice. The currents in Eqs. (26) and (27)
are a function of incidence angle, so for monostatic calculations the integrations must be performed again after
each angle change. Another disadvantage is that modeling the fringe current interaction is difficult. On the
plus side is that the edge-scattered field is simply added to the PO field. This is the major attraction of the
PTD, since PO is so convenient and widely used.
Geometrical Optics. GO is the theory of ray tracing that has been used for centuries to design systems of
lenses such as telescopes. It works well at optical frequencies because the lens and mirror dimensions are much
larger than a wavelength and thus the interaction of the wave with the lens becomes a localized phenomenon.
The reflected ray appears to originate from a single point on the surface, called a specular or reflection point.
GO is the most basic theory that describes wave behavior upon reflection or refraction at an interface
between two materials (10). GO is based on seven assumptions or postulates:
(1) Waves are everywhere locally plane and TEM.
26
RADAR CROSS-SECTION
(2) The wave direction is specified by the normal to the equiphase surfaces, that is, a ray. (Equiphase surfaces
are also referred to as eikonal surfaces.)
(3) Rays travel in straight lines in a homogeneous medium (Fermats principle).
(4) Polarization is constant along a ray.
(5) At the reflection point, the reflected fields are linearly related to the incident fields by reflection coefficients.
(6) Reflected rays satisfy the law of reflection.
(7) Power in a tube of rays is conserved. The field strength is inversely proportional to the square root of the
cross-sectional area of the flux tube.
Ray tubes are also called ray bundles or flux tubes. For a plane wave, all of the rays are parallel and
therefore a bundle of rays has a constant cross section. However, for a spherical wave, a bundle of rays expands
as the distance from the source increases. The power flowing through a slice of the bundle must be constant if
the medium is lossless and power is conserved.
For the general scattering problem the incident wave front and scattering surface may have some curvature, as illustrated in Fig. 12. The local curvature at a point on a doubly curved surface, whether it is a wave
front or target surface, can be expressed in terms of two principal radii of curvature. Let the incident wave
front be characterized by the principal radii of curvature Ri 1 and Ri 2 . Similarly, let the scattering surface be
characterized by the principal radii of curvature Rs 1 and Rs 2 . If these two sets of radii are known, then the
radii of curvature of the reflected wave front can be computed using the following equation:
where f 1,2 are the focal lengths in the two principal planes. (The focal length is another measure of curvature.)
The focal lengths can be computed by (4)
The vectors I1 and t2 are tangent to the surface at the reflection point and lie in principal planes 1 and 2,
respectively.
RADAR CROSS-SECTION
27
Fig. 12. GO can be used to predict the reflected field from a doubly curved surface. The curvatures of the incident wave
front and surface affect the curvature of the reflected wave front.
The postulates allow the reflected field to be computed from the incident field at the reflection point by
multiplying by a reflection coefficient. In matrix form,
where
The subscripts and refer to a polarization where the electric field is parallel or perpendicular to the plane
of incidence (the plane defined by the vectors I and i ).
28
RADAR CROSS-SECTION
The divergence and spreading factor for the reflected wave is
The quantity s is the distance from the source point S to the reflection point Q, and s is the distance along the
reflected ray from the reflection point to the observation point P.
The last factor in Eq. (31) allows for phase transitions through caustics, which are points or lines at which
the cross section of a ray tube is zero (and thus the power density is infinite); we have
A well-known formula for the RCS of a curved surface can be derived using the above formulas. The
fact that for a plane wave Ri 1 = Ri 2 = and the observer is far from the target (Rs 1 , Rs 2 s) yields the
approximation
Equation (37) has been widely applied to basic geometric shapes such as spheres and ellipsoids. Note that GO
predicts zero field in shadow regions. That is, if the path between the source and observation points is blocked,
then there is no received field.
Geometrical Theory of Diffraction. GO provides a means of calculating the field reflected from a surface.
The associated edge effects require a mathematical model for edge diffraction. The GTD is such a ray-based edge
diffraction theory (26). For the GTD, rays are hypothesized that obey diffraction laws similar to reflection laws.
A diffraction coefficient is defined, which depends on the edge geometry and polarization of the incident wave.
Upon diffraction, the scattered field is given by the incident field times the diffraction coefficient. Furthermore,
the diffracted wave follows prescribed straight line paths in free space. The total field at an observation point
is the vector sum of all the reflected and diffracted fields arriving at that point.
As first developed, the GTD had several shortcomings with regard to singularities in the field. Several
modified versions of the GTD, such as the uniform theory of diffraction (UTD) (27) and the asymptotic theory
of diffraction (ATD) (28), have eliminated the problems in many cases by introducing correction factors. In the
following discussions GTD will refer to the collection of these diffraction theories.
The postulates of the GTD are similar to those of GO; only numbers 5 and 6 change:
(1) At the diffraction point, the diffracted fields are linearly related to the incident fields by diffraction coefficients.
(2) Diffracted rays emerge radially from the edge.
In addition, rays traversing a caustic encounter a phase advance of 90 ; an edge is always a caustic. At a
focus there is a phase advance of 180 .
The behavior of diffracted rays from an infinite conducting wedge is illustrated in Fig. 13. The diffracted
rays lie on the surface of a cone (referred to as the Keller cone) with its tip located at the diffraction point and
RADAR CROSS-SECTION
29
Fig. 13. The GTD is used to predict edge diffracted fields. As long as the diffraction point is not near the end of the edge,
the diffracted field is similar to that of an infinitely long edge.
axis along the edge. The cone half angle is equal to the angle that the incident ray makes with the edge,
(a consequence of Fermats principle). The coordinate variables (s, , ) and (s , , ) uniquely define the
directions of diffraction (observation) and incidence (source), respectively.
In the light of postulate 5 the diffracted field from an edge can be written as
The vectors Ed and Ei have the same form as in Eq. (32) and Eq. (33), except that parallel and perpendicular
are defined relative to the edge rather than the plane of incidence. In Eq. (38) D is a two-dimensional matrix
of diffraction coefficients,
30
RADAR CROSS-SECTION
with
In special cases where Ed is tangent to the surface of a PEC, the surface diffracted ray is zero. For a
rectangular plate, there will not be an interaction (higher-order diffraction) between the two parallel edges. In
fact, there is an interaction between two edges of a parallel plate even when the electric field is parallel to the
edge, and it is included by adding a slope diffraction term to Eq. (38)
Rigorous Methods.
Method of Moments. The radiation integrals provide a means of calculating the scattered field from
a target once the induced current on the surface is known. Except for the most simple shapes, the current
distribution on the target is not known accurately. For electrically large bodies that are smooth and sufficiently
flat, the PO approximation provides a good estimate at points far from discontinuities and shadow boundaries.
By a discontinuity is meant any abrupt change in the surface contour or composition. Examples are edges,
steps, cracks, and material joints.
In order to compute the current rigorously, an integral equation (IE) for the current can be formulated
from Maxwells equations and the boundary conditions. IEs have the unknown quantity in the integrand and
possibly in other terms outside of the integral. In most problems of practical interest, the IE must be solved
numerically using a computer. The most popular solution technique, called the method of moments (MM),
consists in expanding the current into a series with unknown coefficients (11). The expansion is inserted into
the integral equation and a weighting process applied. The properties of the expansion functions are used to
reduce the integral equation to a set of simultaneous linear equations that can be solved for the coefficients.
The E-field integral equation (EFIE) is the most commonly used. It can be derived by using the general
form of the radiation integrals and placing the observation point on the surface of the body. [The general form
of the radiation integrals does not make the assumption of a far-field observation point as in Eqs. (17) and
(18).] If the target is a PEC, then (Ei + Es )tan = 0 on the surface. The incident field is known, and the radiation
integrals give the scattered field in terms of the unknown surface current. Thus the following form of the EFIE
RADAR CROSS-SECTION
31
is obtained:
where, as usual, primes denote source point quantities associated with the current and unprimed denotes
observation point quantities. The free space Greens function is defined by
with
Equation (45) is reduced to a matrix equation using the MM. First the current is expanded into a series of basis
functions {J n } with unknown current coefficients {In }:
The basis functions are chosen to be mathematically convenient and adequately represent the current over the
surface. They are of two general types: (1) subdomain or (2) entire domain. In the subdomain case, each function
is nonzero over only a part of the surface. On the other hand, each entire-domain basis function is nonzero over
the entire surface. Using a function of time as an analogy, it can be expressed as a series of sinusoids (Fourier
series), which are an example of entire-domain basis functions. The time function could also be expressed as a
series of steps (a stair step approximation), which is an example of subdomain basis functions.
Once the basis functions are chosen, Eq. (47) is inserted into Eq. (45) and a test is performed that involves
multiplying by a set of weighting functions W n and integrating. Generally the complex conjugates of the basis
functions are selected as the weighting functions (Galerkins method): W n = J n . This leads to N equations
32
RADAR CROSS-SECTION
Fig. 14. Any arbitrary shaped surface can be represented by triangular facets. Triangular (rooftop) basis functions span
each edge. The total current is a superposition of all of the basis functions at a point.
where
which is a voltage. Thus Eq. (48) can be solved for I, then used in Eq. (47), which in turn is used in the radiation
integrals to obtain the scattered field.
A commonly encountered basis function is the triangular subdomain (29). The surface of a target is
represented by a collection of triangular facets. This type of subdomain has the flexibility to model curved
surfaces with arbitrary edge contours. Each basis function is associated with an edge as shown in Fig. 14:
The current at a point in a subdomain is the vector sum of the currents crossing all edges, weighted by the
distance from the edge.
RADAR CROSS-SECTION
33
where
These basis functions result in a linear approximation to the current over each facet. The rigorous result for
the rectangular plate shown in Fig. 11 was computed using the method of moments.
There are a few instances when other basis functions may be more efficient than triangular subdomains.
For example, consider a body of revolution (BOR), which is generated by rotating a plane curve around an
axis. Included in this class are cylinders, cones, spheres, and ellipsoids. A hybrid basis function, that is, a
combination of entire-domain and subdomain basis functions, is an efficient alternative to triangular facets
(30). In these cases overlapping triangular basis functions can be used to represent the current along the
generating curve axis, while entire-domain complex exponentials are used in the azimuthal direction (around
the circumference of the body). The result is a linear approximation to the current in the axial direction (from
back to front) and a Fourier series representation for the current around the body. The major advantage of
this approach is that each Fourier mode, which is actually a matrix whose size is determined by the number of
subdomains along the axis, is independent of all other modes. Thus a large-matrix problem is decomposed into
a series of smaller matrices, thereby permitting larger bodies to be modeled for a given computer resource.
Hybrid Methods. Hybrid computation methods combine several prediction techniques in a single solution (31,32). A hybrid solution decomposes a problem into subproblems, which are solved using different
techniques. Generally the mutual effects (interaction between the subproblems) are included to some degree.
For example, PO is known to give excellent results for large smooth surfaces away from edges and discontinuities. A hybrid approach might use PO on regions of the surface away from discontinuities and the MM near
discontinuities. Larger problems can modeled than with a pure MM approach, and more accuracy is achieved
than using PO separately. The disadvantage is that each hybrid combination is tailored for a specific type of
problem and therefore may not yield accurate results under all conditions.
One hybrid method that is used extensively in RCS calculation for large targets at high frequencies is
the shooting and bouncing ray (SBR) technique (33). It combines GO and PO, and can be supplemented by
34
RADAR CROSS-SECTION
Fig. 15. A complex multilayered target can be replaced by an equivalent target having the same surface contour but an
equivalent surface impedance.
the PTD. Current distributions are determined from the trajectories of rays in a dense bundle. The current
distributions can be used to:
(1) Initiate secondary rays to evaluate multiple reflections
(2) Compute PTD fringe currents for edge diffraction contributions
(3) Compute the far scattered fields
Currently this technique is the most accurate and flexible for computing the RCSs of large complex targets
such as aircraft with engine inlets and cockpits.
Surface Impedance Approximation. The impedance boundary is an approximate boundary condition
that can be used to replace a surface composed of multiple materials and layers by a single surface. It was first
proposed by Leontovich (34) and is frequently referred to as the Leontovich boundary condition. The concept
is illustrated in Fig. 15, which shows the use of a surface current approach to calculate the scattered field from
a target composed of several materials.
The possibility of surface currents flowing on all of the interfaces dramatically increases the mathematical
and computational complexity of the problem over that of a single surface. In many cases it may be sufficient
to replace the collection of surfaces with a single layer that has approximately the same external scattering
properties as the combination of original surfaces. (For the RCS the fields inside the body are not of interest.)
The same scattering properties for the original and equivalent problems are ensured if the boundary conditions
on the tangential field components are the same. The required relationship between the electric and magnetic
fields is referred to as a surface impedance approximation.
Mathematically it relates the tangential components of the electric and magnetic fields at every point on
the surface:
RADAR CROSS-SECTION
35
Comparing Eqs. (53) and (54) reveals that the surface impedance relates the electric and magnetic currents at
every point on the surface:
Note that s = 0 corresponds to a PEC, and the resulting reflection coefficient for a wave normally incident on
the PEC is
where Rs is the surface resistivity in ohms per square. The limiting cases are Rs = 0 for a PEC and Rs = for
a perfectly transparent sheet. The reflection coefficient of the sheet is
36
RADAR CROSS-SECTION
Fig. 16. The RCS of a metal skin aircraft is compared to that of a composite aircraft. The monostatic RCS is lower for
the composite aircraft because more of the incident field is scattered in the forward direction. The surface impedance
approximation has been used.
Resistive films are an effective means of controlling the reflection coefficient of a surface. They consist of
thin metallic deposits on low-dielectric films such as Mylar. A common example of a commercially available
resistive film is window tinting of the type used for automobiles.
The dual of the resistive sheet is the conductive sheet, with the boundary condition
where Gs is the surface conductance. The conductive sheet does not support an electric current. A combination
of resistive and conductive sheets can be viewed as an impedance sheet.
Figure 16 compares the RCSs of PEC and composite aircraft wings. The RCS of the composite wing is
significantly reduced because more of the wave is transmitted and less reflected.
Rough Surfaces. Imperfections and errors in the assembly and manufacturing processes have a significant effect on the low-level RCS. The sources of errors are variations in the physical dimensions and electrical
characteristics of the target materials. Most of the errors behave in a random fashion. Random deviations
in the characteristics of a target result in an increase in diffuse scattering and a corresponding decrease in
specular reflection.
The Rayleigh condition is the traditional criterion for a rough surface:
RADAR CROSS-SECTION
37
where h is the average height of the irregularities and the grazing angle. The surface impedance boundary
condition can also be used to simulate surface irregularities (37).
The calculation of the RCS for targets with random characteristics can be performed using a Monte
Carlo simulation (method of repeated trials) or by applying probability theory to the governing equations
of electromagnetics. Both approaches start by writing the RCS in terms of the random variables of interest
(for example, target dimensions). In the Monte Carlo method, the equations are computer-programmed and
repeatedly executed using a different random seed each time. The average RCS and its standard deviation can
be computed from the output data.
Another approach is to take the expected value of the RCS expression. If the probability density functions
of the random variables are known, then the expectation operation results in an expression for the average
RCS. Using this approach, the RCS of a lumpy surface with a correlation interval c (the distance beyond
which on average the surface deviations become independent) is given approximately by (6)
where
is the variance of the surface deviations (assumed small). P0 is the error-free power pattern
normalized to A2 . (Using a flat plate as an example, P0 = [sinc(ua) sinc(vb) cos ]2 .) The second term in the
brackets of Eq. (60) represents a random noise that increases with the level of the error. A small correlation
interval yields lower, more uniformly dispersed scattering than a large correlation interval.
38
RADAR CROSS-SECTION
is slewed to the target mount. The final result is a plot of RCS versus aspect angle. The data obtained using
the classical far-field CW RCS measurement of a rectangular plate are illustrated in Fig. 17.
For a true far-field measurement the target must be illuminated by a plane wave. This condition is
never rigorously satisfied in practice, because the need for a directive transmit antenna usually results in
a nonuniform-amplitude illumination of the target. Furthermore, the wave incident on the target is always
spherical, not planar. Finally, because the system components are near the floor, ceiling, and walls, reflections
from these surfaces affect the measurement.
Figure 17 illustrates the types of inaccuracies that can occur, such as sidelobe asymmetry and the existence
of a noise floor. The noise floor sets a lower limit on the RCS that can be reliably measured. Measures of chamber
performance are:
(1) Pattern symmetry in angle
(2) Noise floor (noise level)
(3) Quiet zone (linear extent of the region where the amplitude and phase of the incident wave are approximately plane)
(4) Repeatability
(5) Allowable target size and weight for the mount
(6) Allowable range of target motion
(7) Instrumentation capabilities
(8) Data-processing capabilities
(9) Data-presentation capabilities
The state-of-the-art instrumentation used in RCS measurement is essentially a programmable multimode
radar. Data can be collected in either the time or the frequency domain and transformed back and forth between
the two. In the time domain, gating can be employed to discard signals that do not correspond to the time delay
of a scattered signal from the targets extent. Processing such as windowing can be performed on the raw
RADAR CROSS-SECTION
39
Fig. 17. The measured RCS of a 16 inch square plate for a 2 GHz CW source. Errors in the measurement were more
pronounced in older measurement systems. Background subtraction techniques can be used to reduce measurement errors.
data to enhance various characteristics and then see how they behave as a function of frequency or aspect.
Data-processing techniques can even generate high-resolution images that highlight scattering hot spots on
the target.
Chamber Congurations. The majority of all RCS measurement facilities are completely enclosed
(indoor ranges). This provides isolation from the external environment, which is desirable for both security
and technical reasons. In an indoor chamber the interference from other RF sources is reduced or eliminated,
and the temperature is controlled for increased instrumentation stability. The result is a more accurate and
repeatable measurement than possible in an outdoor facility. The obvious limitation on an indoor chamber is
the size of the building required for large targets.
Figure 18 shows two types of indoor chambers. The first is the standard far-field configuration. The
separation of the antenna and target must be large enough so that the difference between an ideal plane wave
and the actual spherical wave is negligible. The required distance is a function of target size and the frequency.
If the distance is R and the target length L, as indicated in Fig. 19, the round trip phase difference between
rays to the edge and center is
40
RADAR CROSS-SECTION
Fig. 18. Common indoor chamber configurations include the conventional far-field range, the tapered chamber, and the
compact range. The compact range is capable of handling physically large targets over a wide frequency range.
where the right-hand side holds for R L. Based on the convention used for the definition of the far field of a
radiating antenna, one may limit the round trip phase error to /8, so that the minimum range becomes
This limit turns out to be overly restrictive and is usually not adhered to in practice. Equation (61) assumes
that all of the target surface contributes coherently to the total RCS, as will happen for example, with a smooth
plate. On the other hand, if the target is composed of many independent scatterers, the far-field distance is
primarily determined by the distribution of dominant scatterers in space. Thus when the dominant scatterers
are clustered, a much smaller range is acceptable (38).
In addition to a plane phase front at the target, the amplitude of the incident wave across the target must
be constant. The amplitude distribution is controlled by the antenna beamwidth and the ratio L/R. An isotropic
antenna provides the flattest phase front, but also results in stronger illumination of the chamber walls and
RADAR CROSS-SECTION
41
Fig. 19. The incident spherical wave from the transmitter deviates from a plane wave. The deviation leads to an error in
the measured RCS. To reduce the error the transmitter to target distance must be increased.
Fig. 20. A compact range provides a uniform field over the target with shorter antenna-to-target distances than a
conventional far-field range. Some sources of measurement error are illustrated.
other obstacles. The transmitted power must also be higher if the antenna gain is low. Thus there is a tradeoff
involved in determining the optimum antenna configuration.
The demands on chamber size required for accurate low-level RCS measurements have resulted in the
development of compact ranges. The layout of a typical compact range is shown in Fig. 20. The parent reflector
configuration is a Cassegrain, and the reflector surfaces have been arranged so that no blockage occurs. The
antenna is located at the focus, and a plane wave field is excited a short distance in front of the paraboloid. The
total volume of a compact range is much less than that of a far-field configuration.
The compact range is not without its own problems, though. To obtain uniform amplitude over the target
requires a shaped feed pattern or some subreflector shaping. Cross polarization and edge diffraction can be
significant. Since the target is close to the reflector edges, the effects of edge scattering are especially important
at low RCS levels. Much effort has gone into the design of rolled or serrated edges that control the level and
direction of edge diffraction.
Sources of Measurement Error. The received signal in a measurement is not entirely due to scattering from the target. It also includes noise introduced by the receiving system and extraneous scattered signals
42
RADAR CROSS-SECTION
Fig. 21. Radar images can be generated by plotting the intensity of the scattered field as a function of the target surface
coordinates. The downrange size of the resolution cell is determined by the radar pulsewidth; the crossrange resolution is
determined by the Doppler measurement capability of the radar.
originating from the chamber walls and other obstacles. Furthermore, some of the transmit signal couples
directly to the receive antenna. This leakage is strongest when a common antenna is used for both transmit
and receive.
Chambers that are used to measure low-level RCSs must be carefully designed, maintained, and calibrated. Great care is taken to reduce the error signals in the vicinity of the target. The cross-sectional dimension
of the chamber that is essentially free of extraneous signals and maintains the plane wave phase and amplitude
is called the quiet zone. It is usually specified in feet or meters. The quiet zone, frequency range of operation,
and noise floor effectively characterize the chamber performance.
Measurement errors are mitigated using background subtraction. First a measurement is taken with the
target absent. Ideally the received signal should be zero if the chamber walls are perfect absorbers and there
is no leakage. In practice a nonzero signal is present due to the sources indicated in Fig. 20, and this value is
stored as a reference. Next, the measurement is repeated with the target present. The measured value includes
the targets RCS as well as the error signals. If the interaction between the target and chamber is negligible,
vector subtraction of the background signal should yield a close approximation to the isolated targets RCS.
The subtraction method requires that the characteristics of the measurement equipment remain stable
during the calibration and measurement runs. If the power level or frequency drifts, the background reference
is no longer valid and the difference is attributed to the target. Even small changes in phase caused by cable
movement can affect low RCS measurements.
Target Mounts. A potential source of large measurement error is the target mount. If at all possible,
the pedestal should be made of low-density material that does not scatter. Frequently the target is suspended
and held in tension using thin polyester filaments similar to fishing line. For large, heavy targets that must
be tilted and rotated there is no alternative but to use metal mounts. In this case special shaped fixtures are
used to shield the pedestal. Absorbing material can also be used in conjunction with shaping to reduce mount
scattering.
Resolved RCS and Target Imaging. Returns from individual scattering elements on a complex target
can be isolated or resolved. The strength of the return is plotted as a function of downrange (y) and crossrange
(x) in Fig. 21. If the cell size becomes small enough, a high-resolution two-dimensional image of the target can
be constructed. As shown in Fig. 22, imaging clearly identifies the strongest scattering sources on a target.
Resolving the individual scattering sources on a target requires that accurate angle and range information be available as the returns are received. Range resolution is achieved by using a short pulse that only
illuminates a small slice of the target. Knowledge of the time elapsed after the leading edge of the pulse hits
the target is equivalent to knowledge of the downrange coordinate y.
One method of obtaining the crossrange value is by accurately pointing a narrow antenna beam. The
beam is scanned over the entire solid angle of the target, thereby providing angle data that can be processed
RADAR CROSS-SECTION
43
Fig. 22. The radar image of the target can be used to determine where RCS treatments should be applied, as well as
evaluate the effectiveness of RCS reduction methods.
Fig. 23. A point rotating on the surface of the target has a Doppler frequency shift. The frequency shift can be used to
determine the crossrange coordinate.
along with the range data to construct an image. This approach is not efficient at microwave frequencies,
however, because of the large antenna size needed for fine angle resolution. It is more practical at millimeter
wavelengths and smaller, and is highly successful at laser wavelengths. A more practical method for obtaining
crossrange exploits the Doppler shift (39). The target is rotated at an angular velocity r as shown in Fig. 23.
Target surface points rotate with a linear velocity that depends on their distance from the center of rotation.
The radial component of relative velocity (component along R) gives rise to a Doppler shift.
Referring to the coordinate system shown in Fig. 23, let R0 be the range to the targets center of rotation
along the y coordinate, and r the distance to a scattering point at a distance d from the center. Using the law
of cosines,
44
RADAR CROSS-SECTION
which simplifies to
when R0 d. The first term under the square root is associated with the round trip time delay to and from the
target center, and the second is the Doppler shift due to rotation. Since = r t (where r is the rotation rate of
the target), the Doppler frequency in hertz is
This method of rotating the target to obtain crossrange information is the same principle used in inverse
synthetic aperture radar (ISAR) method.
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
M. Skolnik, Introduction to Radar Systems, 2nd ed., New York: McGraw-Hill, 1980.
B. Eddie, Radars, Engelwood Cliffs, NJ: Prentice-Hall, 1993.
G. Ruck et al., Radar Cross Section Handbook, New York: Plenum, 1970.
C. Balanis, Advanced Engineering Electromagnetics, San Francisco: Harper & Row, 1982.
K. Kunz, R. Luebbers, The Finite-Difference Time Domain Method for Electromagnetics, Boca Raton, FL: CRC Press,
1993.
D. Jenn, Radar and Laser Cross Section Engineering, Washington, DC: AIAA Education Series, 1995.
A. Ishimaru, Electromagnetic Wave Propagation and Scattering, Englewood Cliffs, NJ: Prentice-Hall, 1991.
L. Peters, End-fire echo area of long, thin bodies, IRE Trans. Antennas Propag., 133, January 1958.
R. Harrington, Time-Harmonic Electromagnetic Fields, New York: McGraw-Hill, 1961.
J. Keller, Determination of reflected and transmitted fields by geometrical optics, J. Opt. Soc. Amer., 40 (1): 48, 1950.
R. Harrington, Field Computation by Moment Methods, New York: Macmillan, 1961.
W. Streetman, J. Goodall, Lockheed F-117A, Osceola, WI: Motorbooks International, 1990.
D. Cheng, Field and Wave Electromagnetics, 2nd ed., New York: McGraw-Hill, 1989.
M. Schwartz, Composite Materials Handbook, 2nd ed., New York: McGraw-Hill, 1992.
V. Weston, Theory of absorbers in scattering, IEEE Trans. Antennas Propag., AP-11: 578, 1963.
N. Engheta, D. Jaggard, Electromagnetic chirality and its applications, IEEE Antennas Propag. Soc. Newslett., 6,
October 1968.
R. Harrington, Theory of loaded scatterers, Proc. IEE Electron., 111 (4): 617623, 1964.
R. Harrington, The control of electromagnetic scattering by impedance loading, Proc. IEEE, 53: 9931004, 1965.
E. Knott, J. Schaeffer, M. Tuley, Radar Cross Section, Norwood, MA: Artech House, 1989.
Navy Times, Jan. 12, 1998, p. 4.
E. Pelton, B. Munk, A streamlined metallic radome, IEEE Trans. Antennas Propag., AP-22: 799, 1974.
D. C. Schleher, Introduction to Electronic Warfare, Norwood, MA: Artech House, 1986.
S. Silver, Microwave Antenna Theory and Design, New York: McGraw-Hill, 1949.
P. Ufimtsev, Method of Edge Waves in the Physical Theory of Diffraction, translated from Russian by the USAF Systems
Command, Foreign Technology Division, Sept. 7, 1971.
A. Michaeli, Equivalent edge currents for arbitrary aspects of observation, IEEE Trans. Antennas Propag., AP-32:
252258, 1984; corrections, AP-33 (2), 1985.
J. Keller, Geometrical theory of diffraction, J. Opt. Soc. Amer., 52 (2): 116, 1962.
R. Kouyoumjian, P. Pathak, A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface,
Proc. IEEE, 62: 1448, 1974.
RADAR CROSS-SECTION
45
28. J. Boersma, Y. Rahmat-Samii, Comparison of two leading theories of edge diffraction with the exact uniform asymptotic
solution, Radio Sci., 15 (6): 1179, 1980.
29. S. Rao, D. Wilton, A. Glisson, Electromagnetic scattering by surfaces of arbitrary shape, IEEE Trans. Antennas Propag.,
AP-30: 409, 1982.
30. L. Medgyesi-Mitschang, J. Putnam, Hybrid solutions for scattering from large bodies of revolution with material
discontinuities and coatings, IEEE Trans. Antennas Propag. AP-32: 717, 1984.
31. E. Ekelman, G. Thiele, A hybrid technique for combining the moment method treatment of wire antennas with the
GTD for curved surfaces, IEEE Trans. Antennas Propag., AP-28: 831, 1980.
32. T. J. Kim, G. Thiele, A hybrid diffraction techniquegeneral theory and applications, IEEE Trans. Antennas Propag.,
AP-30: 888898, 1982.
33. H. Ling, R. Chou, S. Lee, Shooting and bouncing rays: Calculating the RCS of an arbitrarily shaped cavity, IEEE Trans.
Antennas Propag., AP-37: 194, 1989.
34. T. Senior, Impedance boundary conditions for imperfectly conducting surfaces, Appl. Sci. Res., 8(B): 418, 1960.
35. L. Medgyesi-Mitschang, J. Putnam, Integral equation formulations for imperfectly conducting scatterers, IEEE Trans.
Antennas Propag., AP-33: 206, 1985.
36. T. Senior, Combined resistive and conductive sheets, IEEE Trans. Antennas Propag., AP-33: 577, 1985.
37. T. Senior, Impedance boundary conditions for statistically rough surfaces, Appl. Sci. Res., 8(B): 437, 1960.
38. B. Welsh, A minimum range criterion for RCS measurements of a target dominated by point scatterers, 1984 APS/URSI Int. Symp., Vol. APS-17-3, p. 666, 1984.
39. D. Mensa, High Resolution Radar Cross Section Imaging, Norwood, MA: Artech House, 1981.
40. RCS Measurement Capabilities at the Pacific Missile Test Center, Pt. Mugu, CA: Radar Signature Branch.
DAVID C. JENN
Naval Postgraduate School
Power Budget
Power budget (link budget) is the starting point for a radio communication link design. Major factors inuencing
the power budget are shown in Fig. 1. These factors include transmitting and receiving antenna gains GT and
GR , path loss Lp (which depends on operating frequency,
environment and type of radio service), fading margin MF ,
interference margin MI , and radio receiver sensitivity RS.
Transmitting antenna gain (usually specied as the
gain relative to isotropic radiator) and receiving antenna
gain are dependant on the carrier frequency and physical size of the antenna. For a given radio system, selection
of the equipment determines antenna gains. Additional
sources of signal loss at the antenna may include noise due
to protective cover, pointing loss, feeder loss and antenna
efciency described as the ratio of the effective aperture to
the physical aperture of the antenna (5).
Link budget analysis of the radio system denes the
maximum acceptable path loss
L p [dB] = PT [dBm] + GT [dBi] + GR [dBi] MF [dB]
MI [dB] RS[dBm]
which can be used to estimate the coverage area of the
system by determining the maximum transmitter-receiver
distance.
Relationship between path loss and coverage area is
crucial for the design of radio system. Other sources of signal loss in radio propagation are caused by environmental
effects such as rain, clouds, fog, etc.
Fading margin determines the outage probability of the
system due to large-scale and/or small-scale fading. In a
similar way, interference margin can be built into the link
budget to maintain desired performance in the presence
of the interference, regardless of the interference mechanism: adjacent channel interference (ACI), cochannel interference (CCI) in cellular systems, intermodulation distortion (IMD) products created by the large interfering signals, etc.. The amount of margin depends on the required
quality of service. It is expressed in terms of signal to noise
ratio (SNR) for analog systems or bit error rate (BER) for
digital systems, and availability of the link which is the
measure of long-term link utility stated on an average basis, usually on the annual basis.
Radio receiver sensitivity describes the ability of the radio receiver to detect weak signals, and is derived from the
performance requirements of the given radio service. Sensitivity specication includes the impact of techniques that
are integral part of radio reception at the detection level,
such as diversity, coding, interleaving, as well as the impact
of radio receiver such as noise gure and noise bandwidth.
Frequency Planning
Frequency planning is a part of link control and is essential
part of radio link design. Since the radio spectrum is the
limited resource, coexistence of different services will be
practically impossible without the planning procedure. It is
effective way to optimize spectrum usage, enhance channel
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Frequency range
330 KHz
groundwave
ionospheric
High
(HF)
330 MHz
ionospheric
330 GHz
direct waves,
ground
reected
waves
direct waves
30300 GHz
direct wave
frequency
Very
high
frequency (VHF) Ultra high frequency
(VHF)
Super high frequency (SHF)
Radio services
Major characteristics
navigation,
long
distance
radio
telegraphy
navigation,
AM
broadcasting, aeronautical, maritime
comm.
amateur
radio,
broadcasting, aeronautical maritime
comm.
radio and television
broadcasting, land
mobile radio, paging, GPS
satellite,
radars,
short-range communications,
microwave terrestrial
links
very short range
comm, satellite to
satellite links
small bandwidth,
large antennas,
quite large antennas, high power
transmitters
high variability of
channel conditions
relatively
small
antennas, considerable bandwidth
high losses, directional antennas
enormous
bandwidths, rain absorption
kH
0.000154
0.00175
0.0101
0.0367
kV
0.000138
0.00155
0.00887
0.0335
H
0.963
1.308
1.276
1.154
V
0.923
1.265
1.264
1.128
for ACI. This plan is outperformed by N=9 plan with respect to SIR, but the drawback is the reduced channel capacity. Detailed analysis of various frequency plans can be
found in (21). Besides the impact on technical parameters
of the system, frequency planning greatly inuences radio
system growth and economics.
Doppler spectrum
Rice
Class
Class
Class
Class
Class
Doppler spectrum
Class
Class
Class
Class
Class
Class
Doppler spectrum
Class
Class
Class
Class
Class
Class
Table 5. Tapped-delay fading channel models for WCDMA mobile radio: classical Doppler spectrum for all taps (20)
Case 1,speed 3km/h
Relative
Delay
[ns]
0
976
Relative
mean
Power
[dB]
0
10
Relative
Delay
[ns]
0
976
20000
Relative
mean
Power
[dB]
0
0
0
Relative
Delay
[ns]
0
260
521
781
Relative
mean
Power
[dB]
0
3
6
9
Relative
Delay
[ns]
0
976
Relative
mean
Power
[dB]
0
0
Relative
Delay
[ns]
0
976
Relative
mean
Power
[dB]
0
10
Relative
Delay
[ns]
0
260
521
781
Relative
mean
Power
[dB]
0
3
6
9
c
76.8
72.5
67.2
53.6
d
27.7
27.7
27.7
28.6
Table 7. SNR improvement in dB for BER of 1% relative to to single channel reception, comparison of various diversity combining schemes (20)
Diversity branches
2
4
6
Selection
10.0
16.0
18.0
Maximal ratio
11.5
19.0
22.0
Equal gain
10.8
18.0
21.5
for path loss prediction over distances of several kilometres for mobile radio systems and for the line-of-sight microwave links. Assuming that transmitter and receiver antenna heights hT , hR are much smaller than the distance,
path loss is given by
Lground re f l. [dB] = 40 log(d[m]) 10 log(GT [dB])
10 log(GR [dB]) 20 log(hT [m])
20 log(hR [m])
In this case path loss is independent from carrier frequency; however, the inverse fourth power law decay is observed when increasing distance. For distances that are
larger than few tens of kilometres the Earth curvature
should be taken into account and the values of the reection coefcient should be modied (2). Also of importance
is dependence of the reection coefcient on polarization
and materials of the reected surface (2, 3).
Diffraction mechanism allows the radio link to be maintained even when receiver is in the obstructed area, shadowed by the object. Huygens principle could be used to explain the diffraction (2). For practical applications Fresnel
zones should be considered (4). Simple knife-edge diffraction model is often used to calculate signal attenuation.
Objects within Fresnel zones may cause diffraction loss
which is the function of the dimensionless Fresnel-Kirchoff
diffraction parameter
=h
2(d1 + d2 )
d1 d2
L()(dB) =
20 log(0.5 0.62)
20 log[0.5exp(0.95)]
In the cases when there are more than one objects affecting
the radio propagation multiple knife-edge model should be
introduced. Various approximate solutions have been presented, with different level of accuracy. Their comparison
is given in (2).
where f is the carrier frequency, hte is the effective transmitter antenna height, hre is the effective receiver antenna
height, d is the distance and a(hre ) is the correction factor
for effective receiver antenna height which is a function
of the cell size. The correction factor for mobile antenna
height for small to medium size city (urban area) is computed as
a(hre )[dB] = (1.1 log( f [MHz]) 0.7)hre [m]
(1.56 log( f [MHz]) 0.8)
and for a large city (dense urban area) as
a(hre )[dB] =
Suburban area path loss can be calculated using the modied equation
Lsuburban [dB] = Lurban [dB] 2(log( f [MHz]/28))2 5.4
and path loss in open rural areas is given by
Lrural [dB] = Lurban [dB] 4.78(log( f [MHz]))2
18.33 log( f [MHz]) 40.98
The Hata model is well suited for large cell mobile systems, but not for the PCS cells with the radius on the order of 1 km. The COST analysis of several measurements
conducted in European cities for PCS band have resulted
in extended range of parameters for Hatas model to include 1800 MHz frequency band (9). The COST-231 proposed model for path loss is given by
Lurban [dB] = 46.3 + 33.9 log( f [MHz]) 13.82 log(hte [m]) a(hre )
+(44.9 6.55 log(hte [m]))log(d[km]) + CM
where all parameters are dened for original Hatas formula and
CM =
city
and
suburban
areas
k f [dB] = 4 +
0.7( f [MHz]/925 1) f or
1.5( f [MHz]/925 1) f or
Lori [dB] =
and
hmobile = hroof hmobile
hbase = hbase hroof
The multi-screen diffraction loss
Lmsd [dB] = Lbsh + ka + kd log(d[km]) + k f log( f [MHz]) 9 log(b[m])
consists of number of terms:
Lbsh [dB] =
with
ka [dB] =
representing the increase of path loss for base station antennas below the roof tops of the adjacent buildings,
kd [dB] =
18
hbase
18 15
hroof
areas
where
L0 = 35
Molecular Absorption
The absorption from oxygen and water vapour in the atmosphere is the additional factor that has to be considered
at certain frequencies. The atmosphere extends to an altitude of approximately 20 km, yet it represents a path loss
source that cannot be neglected.
Specic attenuation (given in dB/km) is given in a form
of gure in (8). Local maxima of attenuation occur at frequencies around 22 GHz for the water vapour, and 60 and
120 GHz for the oxygen. The respective values are around
0.2 dB/km for water vapour, and around 15 dB/km and 2
dB/km for the oxygen. However, system comparison indicate that water-vapour attenuation is far less important
than the rain attenuation even in the frequency range of
around 20 GHz.
Rain Attenuation
Two components are important when considering the rain
attenuation. First one is polarization related loss due
to non-spherical shape of rain drops. The horizontally
polarized waves experience larger attenuation than the
vertically polarized ones. This can also result in crosspolarization effect that may be harmful for microwave systems using channel planning based on different polarization. The frequency related attenuation depends on the
drop-size distribution, and is not considered signicant for
the signals below 11 GHz for locations with rain climate
similar to northern hemisphere. For tropical areas critical
frequency is as low as 5 GHz.
Rain attenuation statistics prediction is based on rainrate data, which depends on rainfall microstructure. The
ITU-R model is based on rain rate R0.01% exceeded for 0.01%
of the time with an integration time of 1 min. This information can be obtained from local measurements, and in
the case that measurements are not available from ITU-R
reports. The specic attenuation is calculated as
A0.01 [dB] = R r L
Attenuation Due to Clouds and Fog
In Earth-space radio systems for frequencies higher than
10 GHz the attenuation due to clouds may be important
factor affecting the system performance. The mechanism of
attenuation is well understood since clouds and fog consist
of small droplets and it is possible to express attenuation
in terms of total water content per unit volume (12) as
[dB/km] = Kl [dB/km
r=
1
(1 + L/L0 )
g/m3 ] M[g/m3 ]
0.189 f [GHz]
(1 + 2 )
[dB/km/g/m3 ]
where
=
2 +
R = k R
where R is the rain rate in mm/hour, and k, are regression coefcients which are the function of frequency and
polarization (11)
exp(0.015 R0.01 %)
impacts fading characteristics both in case of static transmitter/receiver pair, and or when it is comparable to the
speed of the mobile. Fast fading is observed over distance
of about half a wavelength, with fades of depth about 20 dB
being frequent and larger fades of more than 30 dB being
less frequent, but not uncommon (4).
In practice there is no clear border between slow and
fast fading. It is commonly assumed that in the areas where
multipath occurs radio signal consists of a local mean value
which is constant over a small area, but varies slowly as the
receiver moves, and a fast fading component superimposed
on the slowly varying signal.
In mobile radio communications multipath fading occurs in urban areas where the height of the mobile antenna
is much lower than that of the surrounding buildings and
there is no line-of-sight path between the transmitter and
the receiver. Propagation is mainly determined by scattering and diffraction around the buildings. In xed radio
links where the line of sight propagation exists, multipath
propagation is caused by the reection from the ground or
surrounding objects.
Except for the multipath propagation, other factors may
inuence the fading characteristics. Relative motion between the transmitter and the receiver in mobile and loworbit satellite communications results in a phase change
due to the difference in path lengths, which can be observed
as a Doppler frequency shift in each propagation path as
illustrated in Fig. 5. The phase change is given by
=
2 l
2 vt
=
cos
1
2
v
= cos
t
10
Consequently a multipath fading channel is characterized in statistical terms. The impulse response of the timevarying multipath channel is represented by time-varying
linear lter h(, t) where variable t represents time of transmission and represents the channel multipath delay, for a
xed value of t. Thus h(, t) represents the response of the
channel at time t due to the impulse applied at time t-.
When received signal contains discrete multipath components, the equivalent complex baseband impulse response
model is given by
h(, t) =
where n (t) is the attenuation factor for the n-th path, n (t)
is the propagation delay for the n-th path and f c is the
signal carrier frequency.
Statistical characterization of the fading channel is detailed in (15). In summary, channel impulse response h(, t)
can be represented as a zero-mean complex-valued Gaussian process, so that its envelope at any time instant is
Rayleigh distributed and the phase is uniform (16). In the
2
n n n
mean =
2
n
and rms delay spread is the square root of the second central moment of the power delay prole
=
where
2
2 mean
All delays are measured relative to the rst detectable signal arriving at the receiver at 0 = 0. Also relative amplitudes of multipath components are used in computations.
Typical values of rms delay spread are on the order of microseconds for outdoor mobile radio channels and nanoseconds for indoor radio channels (3).
The maximum excess delay is dened as X 0 , where
X is the maximum delay at which the multipath component is X dB lower than the strongest multipath arrival,
which is not necessarily at 0 . It is important to note that
all parameters highly depend on the noise threshold in the
power delay prole which is used to differentiate multipath
components and thermal noise.
The Fourier transform of the power delay prole determines frequency-coherence properties of the channel. The
inverse of the multipath spread
( f )c =
1
Tm
1
.
50
1
Bd
2 2
2
n n n
=
2
11
0.423
fd
2
2 f d e
e 1
f d 2
12
Channel Classication
The impact of the fading channel on the received signal
depends on the signal characteristics such as signal bandwidth and signal duration. The relationship between signal
characteristics and channel parameters will serve as the
criterion for channel classication. Let us consider digital
communication system with signalling interval T.
For a signalling interval T much larger than the multipath spread of the channel, T > > Tm , channel introduces
negligible amount of intersymbol interference. On the
other hand for a bandwidth of a signalling pulse W 1/T ,
previous relationship imply that the signal bandwidth is
much smaller than the coherence bandwidth of the channel, resulting in frequency-nonselective or frequency-at
fading. This implies that amplitude attenuation and phase
shift introduced by channel are the same for all signal components and consequently spectral characteristics of the
transmitted signal are preserved at the receiver. Channel
distortion is simply a multiplicative random process. In
this case multipath components of the signal cannot be resolved. This case is typical for narrowband communication
signals. Flat fading may result in deep fades.
In a case where signalling interval is smaller than the
multipath spread of the channel, consecutive transmitted
symbols interfere with each other at the receiver, producing
intersymbol interference. In this case signaling bandwidth
W is larger than the coherence bandwidth of the channel,
and different signal frequencies are subject to different
gains and phase shifts. This type of channel is referred
to as the frequency-selective channel. Frequency-selective
channels are typical for wideband communication systems
such as time-division multiple-access systems or spreadspectrum systems.
For a signalling interval T smaller than the coherence
time of the channel, T (t)c , channel characteristics can
be regarded as xed during the symbol interval. In the frequency domain, signaling bandwidth is much larger than
the Doppler spread of the channel. This condition corresponds to a slow fading channel. However, the rate of
change of the channel depends both on signalling bandwidth and the velocity of the transmitter or the receiver.
For a signaling interval larger than the coherence time,
channel impulse response changes within the duration of
the symbol. This results in spectral broadening which in
turn leads to signal distortion since the signal bandwidth
is smaller than the Doppler spread of the channel. This
type of fading is referred to as fast fading or time-selective
fading.
It should be pointed out that there is a slight inconsistency in terminology. Term fast fading is used to describe
multipath fading which is fast compared to shadowing effects (3). On the other hand, multipath fading is characterized as fast fading when signaling interval is larger than
the coherence time of the channel (15). Although it might
be somewhat confusing, precise meaning of the term can
be deducted from the context in which it is used.
For a physical channel with given multipath delay
spread and Doppler spread, different scenarios may arise
depending on the signal design. Slow fading condition is
desirable from the standpoint of receiver design since it
1
=
T
Bd
Tm
S( f ) =
fd
K
1(
f fc
)
fd
xc (t) = 2n = 1 cos(n ) cos(n t) + 2cos cos(d t)
xs (t) = 2n = 1 sin(n ) cos(n t) + 2sin cos(d t)
and Doppler shift is given by d = 2 f d , frequencies of the
sinusoids are n = d cos(2n/N), N0 = (N/2 1)/2, and
phases n = n/N0 and = /4. This model shows excellent agreement with Rayleigh distribution, autocorrelation
Bessel function and Doppler spectrum (12). This technique
can be extended to generate up to N0 independent fading
signals, which is of importance for tapped-delay line fading
channel models for frequency-selective fading.
Tapped-Delay Line Model for Outdoor Mobile Systems.
Tapped-delay line models are commonly used to specify mobile radio channels. The structure of the model is straightforward, and measurements for a particular frequency
band and geographical environment are used to develop
model parameters. Model is specied in terms of number
of multipath arrivals (usually xed), relative delays, relative power of multipath components and Doppler spectrum
for each component. For a given radio standard these parameters may widely vary not only due to different measurements but also due to different signal characteristics
considered.
As an example of tapped-delay line model wideband
propagation model for GSM digital cellular standard is
summarized. Discrete number of taps (6 and 12) is determined by their time delay and their average power, amplitudes of each tap are Rayleigh distributed and varying
according to a Doppler spectrum S( f ) (19). Two types of
Doppler spectrum are dened: classical to be used in all
but one case
S( f ) =
f 2
(1
)
fd
f or
[ f d , f d ]
S( f ) =
2 f d
f or
(1
f 2
)
fd
+ 0.91
( f 0.7 f d )
[ f d , f d ]
Depending on an environment few typical cases are specied: rural area, hilly terrain and typical urban environment. Example of six tap models are summarized in the
Table 4.
With introduction of wider band systems such as Wideband Code Division Multiple Access the model of the propagation channel changes due to different structure of resolvable paths. Moreover, in addition to models where tap delay
is xed as summarized in Table 5, new models with variable
13
tap delay has been introduced (32). The dynamic propagation condition consists of two paths of equal strengths and
equal phases, rst one is static while the time difference
between the two paths is according to
= B +
A
(1 + sin( t))
2
[1
14
of successive impulses, with random amplitudes and random time spacing. Different parameters have been established to qualitatively describe this type of noise including
mean or average voltage, peak voltage, amplitude probability distribution, pulse height distribution, level crossing
rate, pulse duration distribution, etc. (2).
The median value of the man-made noise power is given
as a function of frequency up to 200 MHz by (10)
Fam = c d log( f [MHz])
where c and d take values form Table 6. In the frequency
range 200 MHz to 900 MHz the relationship is given by
Fam = 44.3 12.3 log( f [MHz])
As mentioned earlier, ignition impulses from motor vehicles produce the signicant component of man-made noise,
in particular in VHF band. Noise amplitude distribution
in this case can be determined by
A[dB(V/MHz)] = 106 + 10 logV [vehicles/km2 ] 28 log( f [MHz])
where V is the trafc density.
Predicting the performance of the receiver in the presence of impulse noise is extremely difcult compared to
relatively straightforward method for Gaussian noise. Different techniques are presented in (2).
Cochannel Interference
Cochannel interference is the consequence of better spectrum utilization. It is generated in a scenario where two or
more communication signals are assigned to the same frequency and operate at the same time. CCI may be caused
by a number of scenarios depending on the radio service:
bad route planning for microwave radio links, failure to
carry out co-ordination procedures for the same type of
service, cross-polarization degradation in cross-polar frequency plan, cochannel interference due to frequency reuse
in mobile radio systems, and cochannel interference due to
multiple-access method like in CDMA.
In any case result of the cochannel interference is the
presence of the same-type modulated signal at the receiver
input, which may be lower (in cellular TDMA and microwave systems) or on the same level (in satellite and
cellular CDMA systems) as the desired signal. In most of
the cases cochannel interference problem can be avoided
by careful system planning. This will reduce the cochannel interference to the acceptable level such that system
performance is not signicantly degraded. Technical measures include proper cell design for cellular systems including cell sectorization, power control techniques, and use of
orthogonal codes on the forward link in CDMA cellular systems (note that orthogonal codes in CDMA system can be
used only to eliminate intracell interference, the interference from neighbouring cells will still be present). In the
case when planning methods fail to reduce the amount of
interference, solution may be found in more sophisticated
receiver design using interference suppression techniques.
15
BW
where T is the bit duration (23). Depending on the particular radio system and targeted operating environment, performance measure of the system (SNR or BER) can be evaluated under fading conditions or under static conditions.
In the rst case fading margin is derived from the statistics of the large-scale signal variation such as log-normal
fading. Small-scale signal variations are accounted for implicitly during BER evaluation. When BER of the system is
specied under static conditions, fading margin is derived
from combined statistics of both large-scale and small-scale
variation. Fading margin effectively denes the increase
in SNR required to preserve the same BER as in the static
case.
Radio sensitivity species the lowest detectable signal
level at the radio input; however, both desired and unwanted signals may be present at the radio receiver input at high levels, and it is important to specify the performance for undesired response rejection. The unwanted
strong signal may reduce a nearby weak (desired) signal
resulting in radio desensitization. Desensitization level is
specied for a 1 dB weak signal reduction and is a consequence of radio front-end saturation.
Another factor that may severely limit the performance
of a radio receiver is intermodulation distortion, where
unwanted strong signals produce a number of intermodulation products due to receiver front-end nonlinearities.
IMD products, particularly third-order products, may fall
directly in the bandwidth of the desired signal. To limit the
level of the IMD products receiver is required to have high
intercept point of the particular order.
Dynamic range (DR) of the receiver denes the range of
input signals that a radio receiver can process. Different
considerations may be taken into account depending on
the radio service; however, the most common spurious-free
dynamic range (SFDR) is dened as
SFDR[dB] =
2
(IIP3[dBm] RS[dBm])
3
16
17
plexity constraints.
Selective combining simply selects the diversity branch
with the highest signal level. This type of combining is easy
to implement, additional requirements are antenna switch
and monitoring circuitry. While in selective combining only
one branch is used for subsequent signal processing and
others are discarded, maximal ratio combining performs
coherent combining of all diversity channels. This is optimal method in a sense that provides the maximal signal
to noise ratio for signal detection. The weights applied to
different diversity branches are proportional to the signal
to noise ratios in each of the branches, and can be obtained
either by utilizing pilot signals or sophisticated channel estimation techniques. When variable weighting is not available at the receiver, equal-gain combining may be utilized
by coherently summing the signals from different diversity branches without weighting. This method still allows
the receiver to exploit all diversity channels. Performance
of the equal-gain combining is better than the selection
combining and worse than maximal ratio combining, thus
representing viable technique as a trade-off between performance and complexity. Performance improvement obtained by different combining schemes is summarized in
Table 7.
Diversity reception may be interpreted as the simple
case of coding i.e. repetitive coding since the information
is simply repeated on each diversity channel. Better spectrum utilization can be achieved by employing more sophisticated coding techniques for fading channels, in particular
concatenated coding (15).
All of the diversity reception techniques discussed so
far rely on the processing on the reception. Alternative approach might be to move complexity of signal processing
to the transmitter side, typically base station in the cellular radio system, and preserve simple design of the handset unit while still obtaining diversity gain. These techniques have been commonly termed as transmit diversity
and have been utilized to improve performance in WCDMA
systems (31).
Interference Suppression Techniques
Every radio system operates in the presence of some form
of interference. Regardless of the origin of interference (intentional or unintentional) some form of interference suppression must be employed in order to preserve system performance. Due to the increased number of services, most
of the radio systems in operation are interference-limited
rather than noise-limited. For power budget calculation,
in addition to radio sensitivity which is a function of the
noise bandwidth of the receiver and the required SNR, interference margin should be specied. Alternatively signal
to interference ratio should be specied to satisfy required
BER. Depending on the type of interference (CCI, ACI) different requirements may be imposed on the receiver.
In order to improve the efciency of a radio system different approaches may be pursued: reducing the channel
spacing (increasing ACI), narrowband and wideband system overlay (creating narrowband cochannel interference
to the wideband system), smaller cell design in mobile radio
(increasing CCI), overlapping footprints of the satellites,
18
etc.
Based on the signal observation methods, interference
suppression techniques could be broadly divided into multichannel techniques where multiple sensors (antenna array) are employed for signal reception and single-channel
techniques where only one antenna is employed (25). In
a multichannel scenario, spatial ltering techniques may
realized via adaptive antenna arrays, relying on the different spatial distribution of the interfering signals. Singlechannel techniques rely purely on temporal processing,
usually using adaptive ltering. The most promising path
for the modern receiver design is to incorporate spatiotemporal processing, efciently combining capabilities of
antenna arrays and temporal signal processing (26).
Spatial processing for interference signal suppression
can be carried out using different techniques, including directional antennas, tilted beams and height adjustment.
However, adaptive antenna arrays with capability of adaptive beamforming have shown great potential for congested
radio environments. Different algorithms and optimization
criteria for adaptive beamforming are presented in (27).
Single-channel interference suppression techniques can
be applied both in cases of spread spectrum signalling
and conventional techniques. Although all forms of spread
spectrum signals (direct sequence, frequency hopping, time
hopping) provide certain interference margin due to their
inherent processing gain, additional techniques can be applied to further improve performance. Depending on the
nature of interfering signal, different interference suppression algorithms have been developed. For narrowband
interference rejection in direct sequence systems solutions include adaptive notch lters, decision feedback techniques, adaptive analog-to-digital conversion and other
nonlinear techniques. In the case of CDMA signals, where
interference has the same characteristics as the desired
signal, multi-user detection techniques are developed either in decentralized form (detection of single user of interest) or in centralized form (joint detection of all active
users) (22). Special techniques have been analyzed for frequency hoping systems as well (25).
In nonspread spectrum systems different techniques
have been developed to cope with CCI, ACI and ISI. Adaptive equalization techniques were shown to be efcient
in combating all three types of interference. Other methods include self-optimizing or blind receivers (based on
different techniques like constant modulus, higher-order
statistics, probabilistic approaches etc.), neural network receivers and other nonlinear techniques (25).
Most of the abovementioned interference suppression
algorithms are performed in baseband. In addition to this
different practical interference suppression schemes are
developed at RF/IF stages of the receivers (28).
Signal Losses in Radio Receiver
Due to the inherent complexity of radio receiver number
of signal loss sources are associated with the receiver design. Bandlimiting loss in transmitter and receiver is the
result of nite bandwidth usage, usually restricted by the
radio regulations. Intersymbol interference in which received digital symbols overlap with each other may re-
BIBLIOGRAPHY
1. Rec. ITU-R SM.1047, National spectrum management, 1994.
2. D. Parsons, The Mobile Radio Propagation Channel, New York:
John Wiley & Sons, 1992.
3. T. Rappaport, Wireless Communications Principles and Practices, Upper Saddle River, NJ: Prentice Hall PTR, 1996.
4. S. Sampei, Applications of Digital Wireless Technologies to
Global Wireless Communications, Upper Saddle River, NJ:
Prentice Hall PTR, 1997.
5. B. Sklar, Digital Communications, Englewood Cliffs, NJ: Prentice Hall, 1988.
6. J. Wait, Radiowave propagation, in Wiley Encyclopedia of Electrical and Electronics Engineering, New York: John Wiley and
Sons.
7. W. Lee, Mobile Communications Design Fundamentals, 2nd
ed., New York: John Wiley & Sons, 1993.
8. J. Doble, Introduction to Radio Propagation for Fixed and Mobile Communications, Boston, MA: Artech House Publishers,
1996.
9. COST 231, Urban transmission loss models for mobile radio
in the 900- and 1800- MHz band, COST 231 TD(90) 119, Rev.2,
The Hague, September 1991.
10. Rec. ITU-R, PI.372-6, Radio noise, (1994).
11. Rec.ITU-R, PN.833-1, Specic attenuation model for rain for
use in predicition methods, (1992).
12. Rec. ITU-R PN.840-1,Attenuation due to fog and clouds, (19921994)
13. S. Stein, Fading channel issues in system engineering, IEEE
J.Select. Areas Commun. 5 (2): 6889, 1987.
14. R. Steele, Mobile Radio Communications, London: Pentech
Press, 1992.
ZORAN ZVONAR
Analog Devices, 804 Woburn
Street, Wilmington, MA,
01887
19
187
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
188
Band
Characteristics
Applications
3 Hz3 kHz
ELF
330 kHz
VLF
30300 kHz
300 kHz3 MHz
330 MHz
LF
MF
HF
30300 MHz
VHF
UHF
Long wavelength; inefficient antennas; earth-ionosphere waveguide; penetration of ground and sea
water
Large transmitting antennas; earth-ionosphere
waveguide
Earth-ionosphere waveguide; high atmospheric noise
Ground wave; ionospheric reflections (at night)
Long-distance ionospheric propagation; maximum usable frequency
Line-of-sight propagation; ionosphere scatter; meteor
scatter
Line-of-sight propagation
330 GHz
30300 GHz
SHF
EHF
Medium Frequency
The most popular use of the medium frequency (MF) band is
AM broadcasting, which uses the frequency range from 535
to 1705 kHz. The propagation mode is generally vertically polarized ground wave, although ionospheric reflections (sky
wave) can extend the range at night. Maritime communications is another application.
High Frequency
In the high frequency (HF) band, ionospheric reflections provide the possibility of long-distance communications. The
maximum usable frequency (MUF) for ionospheric reflection
generally occurs in the upper part of the HF band, but the
MUF is a complicated function of the incidence angle and the
state of the ionosphere (5). HF applications include maritime,
aeronautical, amateur, and citizens-band communications.
Very High Frequency
At frequencies above 30 MHz, propagation is predominantly
by line of sight with some refraction (bending) caused by the
atmosphere. However, in some cases ionospheric reflections
can occur at frequencies up to 60 MHz. Scattering from ionized meteor trails (which can be up to 25 km in length) can
be used for meteor burst communications (10). Very high frequency (VHF) applications include television and FM radio
broadcasting and air traffic control and navigation.
Ultra High Frequency
Ultra high frequency (UHF) propagation is essentially by line
of sight with some atmospheric refraction and some scattering by ionospheric irregularities. Applications include UHF
television broadcasting, various radars, satellite communications, and personal communications (11). The global positioning system (GPS) uses transmitting satellites in the UHF
band.
Super High Frequency
Super high frequency (SHF) propagation is primarily line of
sight, and atmospheric absorption becomes significant, partic-
0, 0
189
nents, E and Er, and the radiated magnetic field has only a
single component H . The expressions for these field components can be derived from scalar and vector potentials (12):
1
jk0
1
+ 2 +
sin
r
r
jk0 r3
Il0 jk r 1
1
e 0
cos
Er =
+
2
r2
jk0 r3
1
Il jk r jk0
e 0
+ 2 sin
and H =
4
r
r
E =
Il0 jk r
e 0
4
(1)
(2)
(3)
where the free-space wave number k0 00, the freespace impedance 0 0 / 0, 0 is the magnetic permeability
of free space, and 0 is the dielectric permittivity of free space.
The equations for radiation by the dual source (a magnetic
dipole or small loop) are given in Ref. 12.
Very close (k0r 1) to the Hertzian dipole, the electric field
is dominated by the r3 inverse cube term, and Eqs. (1) and
(2) can be approximated by
E
Il sin
Il cos
and Er
4 j
0 r3
2 j
0 r3
(4)
jk0 Il jk r
e 0 sin and E 0 H
4r
(5)
(6)
where boldface denotes vectors and * denotes complex conjugate. The far-field expression for S can be obtained by substituting Eq. (5) into Eq. (6):
Il
S rr
0
x
Figure 1. Geometry for radiation from a short electric dipole.
k0 |I|l sin
4r
2
(7)
190
P=
d
0
d r2 sin r S =
0 (k0 |I|l)2
6
(8)
The result for the total radiated power in Eq. (8) is independent of the radius r at which the integration is evaluated, but
the evaluation is simplest in the far field, where S can be
approximated by Eq. (7).
is incident on a half space with permittivity , electrical conductivity , and permeability . The angle of reflection r
equals the angle of incidence, r 0. The reflection coefficient
depends on the polarization of the incident field.
For horizontal polarization (electric field perpendicular to
the plane of incidence), the reflection coefficient h is given by
(1)
k0 cos 0 0 k2 k20 sin2 0
h =
k0 cos 0 + 0 k2 k20 sin2 0
(9)
where Pt is the transmitted power, Gt is the gain of the transmitting antenna, Gr is the gain of the receiving antenna, D is
the separation distance between the antennas, and the freespace wavelength 0 2/k0. The D2 factor represents the
inverse-square dependence of the radiated power density
available at the receiving antenna.
The antenna and propagation effects can be separated by
writing Eq. (9) in the following form:
2
0
Pr
= Gt Gr
(10)
Pt
4D
The squared factor on the right side of Eq. (10) is dimensionless and does not involve the antenna gains. The reciprocal is called the free-space transmission loss L0 and is usually
expressed in decibels:
2
4D
4D
L0 = 10 log10
dB = 20 log10
dB
(11)
0
0
Normally the antenna gains and the power ratio in Eq. (10)
are also expressed in decibels.
0 k2 cos 0 k0 k2 k20 sin2 0
v =
0 k2 cos 0 + k0 k2 k20 sin2 0
h =
v =
;;;
Reflected
field
r = 0
0 , 0
,,
cos 0
cos 0 +
(k/k0 )2 sin2 0
(k/k0 )2 sin2 0
(14)
and
Plane-Wave Incidence
Incident
field
(13)
(12)
(k/k0 )2 cos 0
(k/k0 )2 cos 0 +
(k/k0 )2 sin2 0
(k/k0 )2 sin2 0
(15)
r sin2 0
h =
cos 0 +
r sin2 0
(16)
r sin2 0
v =
r cos 0 +
r sin2 0
(17)
cos 0
and
r cos 0
where r / 0.
As incidence angle 0 varies from 0 (normal incidence) to
/2 (grazing incidence), h varies smoothly from (1 r)/
(1 r) to 1. However, for vertical polarization, the reflection coefficient v equals 0 at the Brewster angle, B
tan1(r). At this angle, all of the incident energy is refracted
into the dielectric. An examination of Eq. (15) reveals that the
presence of nonzero conductivity (which yields a complex
k) prevents v from going to zero. However, if the imaginary
e
Dir
ct
ray
a
dr
y,
;;;
fl
Re
ec
te
E | = =
Il
,,
0 , 0
Dipole Sources
j0 Il sin jk r jk h cos
e 0 (e 0
+ v ejk0 h cos )
4r
(18)
where v is given by Eq. (13). For the special case of a perfectly conducting ground plane ( ), the reflection coefficient equals 1, and Eq. (18) reduces to
E | =0 =
j0 Il sin jk r
e 0 cos(k0 h cos )
2r
(19)
191
j0 k20 IA sin jk r
e 0 sin(k0 h cos )
2r
(21)
(22)
0
or t = sin1
sin 0
n
(23)
(20)
;;;;
;;;
;;;
where h is given by Eq. (12). For the special case of a perfectly conducting ground plane ( ), the reflection coeffi-
Incident
field
Reflected
field
e
Dir
ct
ray
Re
fle
r
ted
ay
,h
0 , 0
, 0
IA
0 , 0
,,
Transmitted
field
192
E t = T exp(jkz), where k = (
j /)
(28)
2 cos 0
Th =
cos 0 +
r sin2 0
(24)
2
r cos 0
Tv =
r cos 0 +
r sin2 0
0 =0
exp(jk0 z)
;;;
;;;
(27)
0 , 0
,,
Transmitted field
Z=0
Reflected field
Transmitted field
2
+ 0
(30)
(31)
1
=
=
Im(k)
(29)
(25)
Lossy Medium
E r = h |
2k0
k0 + 0 k
T=
E i = exp(jk0 z)
T=
, where a = 2
and b =
a2 + b 2 a
(32)
2
=
1
f
(33)
;
;
;;
;;;
a
Air
193
(d,-h)
Observer
IA
Q
2h3
(34)
where
Earth
Hz =
Q=
0
2
2
2J (ga/h)
g3 e g +jH
dg
J (gD) 1
2 0
ga/h
2
g + g + jH
(35)
H 0h, D d/h, and J0 and J1 are the zero- and firstorder Bessel functions (17). The integration variable g is a
normalized wave number; so the integration in Eq. (35) can
be interpreted as the result of a superposition of all the waves
that the loop antenna transmits into the earth. The exponential factor is consistent with skin depth attenuation because
for small g it is given by exp(jH2) exp(h/ ), where
is given by Eq. (33).
When a/h 1, the factor 2J1(ga/h)/(ga/h) approaches 1
over the significant range of g. In that case the dependence
on a enters only on through the loop area A(a2), and the
loop radiates like a magnetic dipole of moment IA. If the loop
is buried at a depth h and the observer is located at the surface (as in uplink communications), the result for Hz is identical.
For the special case of H a/h 0, the quantity Q reduces to the following result for a static magnetic dipole:
Q|H=a/h=0 =
2 D2
2(1 + D2 )5/2
(36)
194
1.0
1.0
0.5
0.2
0.1
D=0
a/h = 0
0.05
0.02
0.01
2
4
6
8
Normalized depth (H = 0 h)
10
a/h = 0
0.6
a/h = 0.5
0.4
a/h = 1
0.2
a/h = 2
0
1
H=0
0.8
1.0
H=0
0.8
1
0.6
2
z
y
0.4
Air
0.2
Earth
(x , 0, 0)
x
l
l
R
0
0
Observer
(x, y ,z)
Figure 11. Electric line source on a conducting half space with a
subsurface observer.
field is received with a loop antenna, or the subsurface electric field is received with a grounded wire antenna.
First consider the fields produced by an incremental current source of length dx located at x, as shown in Fig. 11.
Because of the quasi-static approximation, the Sommerfeld
integral forms for an incremental source of current moment
I dx can be greatly simplified (21). As a result, the magnetic
field components are
P
N
x y z2
x y z
3
4N
I dx
P
3P
+
dH y =
+
2 2 z3
x2 z z2 y2
dH x =
I dx
2 2
and
dH z =
Idx
2 2
2N
3P
4N
2
3
y z
y z y z2
(39)
I dx
2
2P
3N
+ 2
2
z
y z
h
2
l
l
F (H, Y, X , L) =
4N
3P
3P
+ 2 2
+ 2
3
z
x z z y
2N
3P
4N
2
3
y z
y z y z2
dx
(46)
dx (47)
1
2
l
l
3N
2P
+ 2
2
z
y z
dx
(48)
(40)
0.8
(41)
Y=0
(42)
I
A(H, Y, X , L)
2h
(43)
Hz =
I
B(H, Y, X , L)
2h
(44)
0.7
and
I dx 2P
dEz =
2 x z
I dx 3N
dEy =
2 y x z
h
2
and
(37)
(38)
195
Y = 0.3
0.6
0.5
Y = 0.7
0.4
0.3
H = 2,
X=0
0.2
Y=2
0.1
and
j0 I
F (H, Y, X , L)
Ex =
2
0
0
(45)
196
pendence of the subsurface electric fields. The frequency dependence of all three components of the electric field has been
analyzed (23), but only the dominant component Ex, will be
considered here. Equations (45) and (48) can be recast in the
following equivalent form:
0.25
Y = 0.7
0.20
Ex =
Y = 0.3
0.15
I
Exn (W, L, X , Y )
2 h2
(49)
where
Exn = h2
0.10
3N
2P
+ 2
2
z
y x
dx
(50)
Y=2
0.05
H = 2, X = 0
dc
3
Exn
= (L X )R3
1 + (L + X )R2
(51)
where
(11)(14) and other calculations (21) shows that this is approximately achieved for L greater than about 2. This means
that the antenna length 2l should be at least four times the
depth h of interest in a particular geophysical application.
Straight, grounded wires are also used to excite broadband, transient fields in the earth. To illustrate the dispersive
nature of the earth, it is useful to examine the frequency de-
Y=0
0.30
(X L)2 + Y 2 + 1 and R2 =
(X + L)2 + Y 2 + 1 (52)
Y = 0.3
0.25
0.35
R1 =
Y = 0.7
0.20
0.15
H = 2, X = 0
0.10
Y=2
0.05
0
0
1.4
L=1
X = 0, Y = 0
1.2
1.0
0.8
X = 1,Y = 0
0.6
0.4
X = 1,Y = 1
0.2
0
0.1
1.0
10
100
;;;
z
z=0
Zs
air
Layer 1
o, 1, 1,
Il
197
Ez
Ez =
j0 Il jk
e 0 F (p)
2
(56)
hs
Layer 2
o, 2, 2,
Figure 16. Vertical electric dipole source at the surface of a twolayer half space. The surface impedance Zs depends on the properties
of both layers.
GROUND-WAVE PROPAGATION
When both the transmitting and receiving antennas are located on the ground, the direct and reflected waves cancel.
Consequently, the inverse-distance space wave is 0 as indicated in Eq. (22). In this case the ground wave is the dominant field component, and it can be calculated for either a
flat-earth model or a curved-earth model.
where
F (p) = 1 j pe p erfc ( j p )
(57)
1
, 3/2 phase (p) 0
2p
(58)
or
Flat Earth
The flat-earth model is useful for analyzing propagation along
flat surfaces (25) or curved surfaces with very large radii of
curvature (such as the earth). Consider a vertical electric dipole (VED) source located at an impedance surface, as shown
in Fig. 16. The air region has free-space permittivity 0, and
free-space permeability 0 is assumed everywhere. The nonzero field components are Ez, E and H .
For propagation analysis, the earth can be characterized
by a surface impedance Zs under fairly general conditions.
The surface impedance is the ratio of the horizontal electric
and magnetic fields at the earth surface: Zs (E /H )z0.
For the two-layer model shown in Fig. 16, the surface impedance for grazing incidence (as in ground-wave propagation) is
given by (26)
Zs = Z1 Q
(53)
where
Z1 = 1 1
Q=
12
02
0
, u1 = 12 + k20 , u2 = 22 + k20 , 12
1 j1 /
= j0 (1 + j
1 )
1 =
(54)
1
, 0 < phase (p) /2
F(p) 2 j pe p
2p
(59)
0 Il
ejk0
2 2 2
(60)
;;;;
(55)
Il
Zs
Ez
Figure 17. Elevated source and observer for propagation over a flat
surface with surface impedance Zs.
198
j0 Il jk
e
F (p)G(h)G(z)
2
(61)
where
G(h) = 1 + jk0 h and G(z) = 1 + jk0 z
(62)
j0 Il jk d
e 0 W
2d
(63)
Il
x
exp( jxts )
j s=1 ts q2
(64)
d
Er
Zs
(65)
Am (e j /4 qx1/2 )m
(67)
m=0
where
A0 = 1, A1 = j 1/2 , A2 = 2
1
1
4
A3 = j 1/2 1 + 3 , A4 =
1+ 3 ,
4q
3
2q
3
1
j 1/2
8
7
1 + 3 , A6 =
1+ 3 +
A5 =
2
4q
15
q
32q6
1/2
5
3
j
1
16
27
1 + 3 + 6 , A8 =
1+ 3 +
A7 =
6
4q
2q
105
2q
32q6
1/2
7
j
5
21
1+ 3 + 6 +
A9 =
24
4q
4q
64q9
and
A10 =
64
32
11
7
+
+
+
3
6
945 945q
189q
270q9
In the numerical results to follow, the series has been truncated at m 10 because higher-order coefficients Am are not
available in the literature.
The small-curvature expansion for W is given by (31,32)
1
W = F (p) + 3 [1 j p (1 + 2p)F(p)]
4q
2
p
1
5p2
+
1 F (p)
+ 6 1 j p(1 p) 2p +
4q
6
2
(68)
102
Phase ( ) = 90
89
101
85
100
60
101
0
15
30
102
= 0.1
f = 10 MHz
103
0.1
10
1
5
11
539
1
+
+
+
+
2q 8q4
32q7
32q10
512q13
4
7
31
397
j2q2 exp q3 1
3
12q3
48q6
288q9
t0 q 2 +
f = 10 MHz
x
exp(jxts )
Gs (y0 )Gs (y)
j s=1 ts q2
Gs (y) =
30
0.1
w(ts y)
w(ts )
(71)
Residue series
Small curvature
Power series
Flat earth
75
104
(70)
where
101
103
(69)
Part of the reason for interest in inductive surface impedances is in modeling propagation over a layer of sea ice on the
ocean (29).
The residue series in Eq. (64) can be generalized to the
case of an elevated source and an elevated observer. For a
source height of h0 and an observer height of h, W takes the
following form (28,29):
Phase ( ) = 30
102
100
d (km)
= 0.1
45
W=
100
75
101
Phase ( ) = 75
199
10
100
Gs (y) = 1 qy +
300
d (km)
Figure 19. Magnitude of the spherical earth attenuation function as
computed by various methods. (Results are shown outside their regions of validity for comparison purposes.) From Ref. 27.
ts 2 1 + ts q 3
y
y +
2
6
(72)
(73)
200
o, o
d
Il
h
Zs
Observer
d1
h1
Zs1
j0 Il jk d
e 0 W (x, q, q1 )
2d
sents no difficulty. A useful alternative to Eq. (75) can be obtained by reciprocity (41):
x
W (x, q, q1 ) = W (x, q1 ) +
(q q1 )
j
xx
1 W (x x,
q1 )W (x,
q)
dx (77)
x(
x x)
0
Equations (75) and (77) have been used to calculate W for
a variety of paths and parameters (42). Results for W are
shown in Fig. 22 for propagation along a land-to-sea path for
various lengths of the land section. The frequency is 10 MHz,
the land constants are / 0 15 and 0.01 S/m, and the
sea constants are 1 / 0 80 and 1 4 S/m. The increase in
field strength that occurs in crossing the land-sea boundary
has been called the recovery effect, and it has been observed
experimentally by Millington (43). A similar drop in phase
has been calculated (42) and observed experimentally by
Pressey et al. (44). In Fig. 23, the length of the land section
is fixed at 5 km, and the frequency is varied from 1 to 30
MHz. The recovery effect is most prominent at 30 MHz because the attenuation over land is so rapid at that frequency.
When either height, h or h1, is nonzero, Eqs. (75) and (77)
are still valid, but W needs to be modified by the appropriate
height-gain function:
x
exp(jxts ) w1 (ts y)
,
W (x, q) =
j s=1 ts q2
w1 (ts )
W (x, q1 ) =
(79)
(74)
All sea
d d1 = 0.4 km
1 km
2 km
5 km
101
10 km
20 km
Al
l
la
nd
x
(q q)
j 1
x1
W (x x,
q)W (x,
q1 )
dx,
d1 > 0
x(
x x)
0
x
exp(jxts ) w1 (ts y)
,
j s=1 ts q21
w1 (ts )
W (x, q, q1 ) = W (x, q) +
(78)
(75)
50 km
102
where
W (x, q) =
x
exp(jxts )
j s=1 ts q2
(76)
x (k0a/2) (d/a), and x1 (k0a/2) (d1 /a). The sphericalearth attenuation function in Eq. (76) is the same as that for
the uniform path in Eq. (64), but the arguments are shown
explicitly for use in Eq. (75). The square-root singularity in
Eq. (75) is integrable, and its numerical evaluation (42) pre1/3
1/3
103
0.1
1.0
10
100
d (km)
201
2
1 MHz
100
3 MHz
101
200m
10 MHz
102
100m
f = 30 MHz
50m
101
h1 = 0
103
0.1
1.0
10
100m
100
d (km)
W (x, q, q1 ) =
x
(q1 q)
j
s=1 t=1
(80)
(81)
The double summation in Eq. (80) converges slowly when either x1 or x x1 is small, but the roots ts and t(1)
r can be computed rapidly (27).
Results for W are shown in Fig. 24 for propagation along
a land-to-sea path for various observer heights. The length of
the land section is 10 km. As many as 600 roots, ts and t(1)
r ,
were used in computing the results (42) in Fig. 24. A good
confirmation of the validity of the numerical results in Fig. 24
as computed by Eq. (80) is that the curve for h1 0 agrees
with the corresponding curve in Fig. 22 as computed by Eq.
(75). As the observer height h1 is increased, the recovery effect
decreases in magnitude and begins at a range beyond the
land-sea boundary.
Irregular Terrain
For many propagation paths, the ground parameters and terrain height vary as a function of position along the path. Analytical methods are not general enough to handle such variations, but the integral equation approach (45,46) has been
found useful for ground-wave propagation over irregular, inhomogeneous terrain. Otts program WAGNER has been thoroughly tested and used on a wide variety of paths (47), and
102
1
10
d (km)
100
Receiver
0, 0
Il
ha
hr
Origin
y(x)
x
Slab
Integration
point
y( )
D( )
Ground
x
Figure 25. Geometry for the integral equation solution of propagation over irregular, inhomogeneous terrain. The slab and ground parameters can vary as a function of x, but are constant in y.
202
(82)
(86)
where f(x) is the attenuation to be determined from the integral equation. The integral equation is (47,48)
Gs (h) = 1 + jk0 h, h 0
(87)
jk0 x
f (x) = F(x, 0)
f ( )ejk0 (x, )
where is evaluated at the appropriate source or receiver
2 0
location. For the source or receiver in the slab, the heighty(x) y( )
+ [( ) a ]F (x, )
y ( )F(x, )
gain function is
x
x
1 ev 0 z + Rev 0 (2D+h)
d (83)
Gs (h) =
, D < h < 0
(88)
(x )
vc
1 + Re2v 0 D
where
y ( ) y (x)
[y(x) y( )]
+
2x(x )
2
2x
and y() is the slope dy/d. The normalized surface impedance () is a function of the slab and ground parameters (48):
( ) = 1
2 + 1 tanh(v0 D)
1 + 2 tanh(v0 D)
(84)
where
gc 1
hc
, 2 =
, v0 = jk0
hc
hc
gc
=
h + h /( j
0 ),
gc =
g + g /( j
0 ),
1 =
hc
vc =
v + v /( j
0 ), =
hc /
vc
v and v are the vertical conductivity and relative permittivity of the slab, h and h are the horizontal conductivity and
relative permittivity of the slab, and g and g are the conductivity and permittivity of the ground. The ground and the slab
have free-space permeability 0. The normalized surface impedance a is evaluated at the source.
Before discussing the numerical solution of Eq. (83), it is
useful to examine the special case of a uniform path: y()
y(x) y() 0 and () a. Thus the integrand is 0, and
f(x) is
(85)
where pa jk0a2x/2. Equation (85) agrees with the flatearth result in Eq. (56). When either y or varies along the
path, the integral equation must be solved numerically. A forward-stepping solution (49) in x is used to obtain values of f
at discrete values of x along the path. Since Eq. (83) is a Volterra integral equation of the second kind, the value of f(x)
depends only on the previously computed values of f() for
x. Physically this means that backscatter is neglected.
To account for nonzero source and receiver heights, ha and
hr, an attenuation function f h is determined by use of height-
(89)
(90)
203
100
Transmitter
Receiver
Gd = 20 log10 |F (v)|
Forest
Ground
(92)
101
fh
102
BIBLIOGRAPHY
Aperture
theory
Integral equation
103
0
Distance (km)
Figure 26. Propagation from a clearing to a forest at a frequency of
10 MHz. From Ref. 48.
Knife
edge
d1
S
h
d2
R
204
DAVID A. HILL
National Institute of Standards and
Technology
204
RADIOWAVE PROPAGATION IN
MULTIPATH CHANNELS
With the rapid growth in wireless communications, there is
considerable interest in radiofrequency (RF) propagation and
RF channel modeling in the 400 MHz to 2 GHz band, where
multipath propagation is a characteristic property. The topic
of RF signal propagation and the factors that affect the RF
signals have been extensively studied over the last few decades (13). This article presents a comprehensive overview
of the RF signal variations related to propagation in
multipath fading channels. The diverse phenomena that
cause signal variations are described via mathematical models. The different types of fading and their salient features
are discussed in detail. The goal of this article is to provide a
mathematical and an engineering-oriented treatment of
multipath fading, thereby providing the reader with the necessary tools and the information to understand the different
RF propagative issues and how they impact wireless communications.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
The different propagative effects are classified in two major categories as shown in Fig. 1:
small-scale effects
large-scale effects
The small-scale effects are those caused by multipath propagation, the phenomenon by which the transmitted signal
reaches the receiver via multiple paths. The received signal
is a superposition of the different multipath signals, which
add constructively or destructively, thereby producing signal
fluctuations. The variations of the signal level are typically
produced by movement over short distances (1020 times the
wavelength of the RF carrier) during which the mean signal
level remains constant. The different types of fading and their
mathematical characterization and modeling are discussed in
the following sections.
On the other hand, large-scale effects are those caused by
variations in the distance between transmitter and receiver,
which are manifested in the form of path loss, and by environmental factors, such as terrain, which produce log-normal
shadowing about the local mean signal level. These issues are
also discussed in detail.
A schematic of a communication system is shown in Fig. 2.
The term channel is used to imply different portions of the
communication link. Two terms used in this article, namely,
the propagative channel and the baseband channel need to be
defined. The propagation channel is the physical medium that
supports RF propagation between the transmitting and receiving antennas. This includes the objects that cause the
BASIC DEFINITIONS
Static Channels
The simplest type of communication channel is a static channel, in which the only signal impairment caused by the channel is additive white Gaussian noise (AWGN). The noise is
assumed to have a flat, power-spectral density and is uncorrelated in time. These characteristics are typical of the thermal
noise generated in communication equipment.
The AWGN channel is widely used to evaluate the performance of communication systems (4) (modulation, coding,
etc.). In RF communications, a static environment is unlikely,
except in a few situations when there is a dominant line-ofsight (LOS) component between the transmitter and receiver,
and there is essentially no multipath, such as in microwave
LOS links and in microcells. In practice, the AGWN channel
serves the useful role of providing performance bounds for RF
communication channels because the practical channels (typically Rayleigh and Ricean) always produce signal impairments more severe than the AGWN channel.
Signal variations
due to propagation
Small-scale effects
Large-scale effects
Multipath fading
Free space
Rayleigh fading
Propagative loss
Indoor
Ricean fading
Outdoor
Flat fading
Reflection
Freq.-selective fading
Propagative modes
Diffraction
Fast fading
Scattering
Slow fading
Time dispersion
205
Shadowing
Random FM
Figure 1. Classification of the small-scale effects and the large-scale effects that cause signal
variability in an RF environment.
206
Propagative
channel
Input
bits
Encoder
Modulation
RF/IF
stages
RF/IF
stages
Demodulation
Outputs
bits
Decoder
Fading Channels
Some of the commonly encountered communication channels,
such as that encountered in wireline communications over
telephone networks, can be characterized as AWGN channels
that have a linear time-invariant (LTI) filter channel model.
However, this representation does not apply to radio frequency (RF) channels because their transmission characteristics change with time and hence require statistical characterization using linear, time-varying channel models. Typically
RF channels are characterized by the following time-varying
phenomena:
1.
2.
3.
4.
5.
can be expressed as
x(t) =
n (t)s[t n (t)]
(1)
where s(t) is the modulated signal, n(t) and n(t) are the attenuation factor and the propagative delay, respectively, for
the signal received via the nth path. The modulated signal
s(t) is typically represented as (4)
s(t) = m(t) cos[2 f ct + (t)] = Re[u(t)e j2 f c t ]
(2)
x(t) = Re
n (t)e
j2 f c n (t )
u[t n (t)] e
j2 f c t
(3)
Transmitted pulse
Time
Received pulses
Time
Figure 4. Effect of time dispersion due to multipath propagation.
207
c( ; t) =
n (t)e j2 f c n (t ) [t n (t)]
(5)
which also represents the response of the channel to a complex exponential ej2fct. Even though s(t) is a monochromatic
signal, the output of the channel r(t) contains many different
frequency components, which are generated as a result of the
time variations of the channel response. The range of frequencies over which the spectrum of r(t) is non-zero is called the
Doppler frequency spread (Bd) of the channel, which yields a
measure that is directly proportional to the rate of variations
of the channel response.
Signal Fading. In Eq. (5), c( ;t) is the time-varying (complex-valued) channel impulse response (CIR) and is the discrete multipath channel model, wherein represents the time
dispersion and t denotes the time index for the variations as
a function of time. The CIR c( ; t) can be viewed as the superposition of phasors, each with a time-varying gain/amplitude
n(t) and a time-varying phase n(t) 2f cn(t). With random variations in n(t) and n(t) (caused by the time-varying
multipath), these phasors can add constructively or destructively. The fading phenomenon, which results in signal amplitude becoming very small, is caused by instantaneous destructive combining of the phasors. An important observation
is that a small change in n(t) produces a significant phase
0
5
10
15
20
25
30
35
40
0.2
208
FADING DISTRIBUTIONS
The two most commonly encountered types of fading environments are
1. Rayleigh fading
2. Ricean fading
In this section, the statistical properties of Rayleigh and Ricean fading are described. This mathematical framework is
very useful in providing tools for analyzing and simulating
the performance of communications over fading channels and
also for providing insight into the design of robust communication systems in a fading environment (see Ref. 16). Ricean
fading covers a very broad spectrum of environments from
static AWGN channels on the one end to Rayleigh fading
channels on the other. So, it represents an entire family of
fading channels. However, because Rayleigh channels are the
most frequently encountered environment in wireless (mobile) communications, it forms an important class for study.
Both Rayleigh and Ricean distributions refer to the statistics
of the received signals envelope.
RICE DISTRIBUTION
Let X and Y be two independent Gaussian random variables
(RVs) with means m1 and m2, respectively, and common vari
X2 Y2, is distribance 2. Then the RV V, defined as V
uted according to Ricean distribution (8), whose pdf is given
by
2 2
v I sv e v 2+s2
0
2
2
fV (v) =
v>0
(8)
v0
I0 (x) =
1
2
ex cos u du =
0
1
2
ex cos u du
(9)
V
0
fV (v) dv = 1 Q
s v
,
(10)
a 2 +b 2
2
k
a
k=0
Ik (ab),
b>a>0
Rayleigh Distribution
0.7
As shown in the preceding section, the Rayleigh distribution is a special case of the Rice distribution (K 0 dB). The
Rayleigh distribution can also be derived independently from
the central chi-square distribution, as shown herein. Let X
and Y be two independent identically distributed (iid) zeromean Gaussian RVs with variance 2. Then the RV V defined
Rayleigh
0.6
pdf of RV fv(v)
0.5
Rice, K = 1
Rice, K = 4
0.4
Rice, K = 10
0.3
v e 2v 2
2
fV (v) =
0.2
0.1
0
209
Amplitude of RV V
Figure 6. Plot of the Rayleigh pdf and the Ricean pdf for different
values of the Rice factor K 1, 4, and 10.
(11)
v<0
v0
V
0
fV (v) dv = 1 e
2
v 2
2
,v 0
(12)
and
E[V 2 ] = 2 2
10
0
10
20
30
0
Rayleigh
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
Time (s)
0.12
0.14
0.16
0.18
0.2
10
0
10
20
30
0
Rice, K = 1
0.02
0.04
10
0
10
20
30
0
Rice, K = 4
0.02
0.04
10
0
10
20
Rice, K = 10
30
0
0.02
0.04
210
k
E[V k ] = (2 2 )k/2 1 +
2
and
2 = E[2 ] 2
2 m m 2m1 mv 2
v
e , m 1/2
(m)
(13)
where E[V2], (m) is the gamma function, and the parameter m is called the fading figure, defined as
m=
2
,
)2 ]
E[(V 2
m 1/2
v
cos n (Hz)
(15)
(K + 1)2
(2K + 1)
Log-Normal Distribution
The variations in the local mean of the signal power received
in a fading environment is referred to as shadowing. These
variations are caused by terrain and other environmental factors. Shadowing is characterized via the log-normal distribution, as defined herein. An RV is said to have a log-normal
distribution if its pdf is given by
1
f () =
e
2
( ) 2
2 2
(14)
n
where
= 20 log10
= E[]
Mobile direction
(velocity v )
Figure 8. Reference figure for deriving the mathematical model for
Rayleigh fading.
Ez = E0
N
Cn cos(2 f ct + n )
(16)
n=1
(17)
and
s
sin =
211
f f 2
c
(21)
fm
3b
1
s
2 f m
2 , | f f c | f m
f fc
SE z ( f ) =
(22)
1
fm
0
| f fc| > fm
where b is a scale factor denoting the mean power received
by an isotropic antenna. A normalized plot of the power spectrum is shown in Fig. 9. The peak of the power spectrum occurs at f m. The power spectra for the magnetic field components are given in Ref. 1.
Envelope Correlation for Rayleigh Fading
where
Tc (t) = E0
N
Cn cos(2 f n t + n )
n=1
Ts (t) = E0
N
Cn sin(2 f nt + n )
(18)
n=1
Both Tc(t) and Ts(t) are Gaussian random processes. The corresponding RVs Tc and Ts have zero mean and equal variance
given by
=
fc fm
fc + fm
fc fm
3b
2 f m
Tc2
= Ts2
= E02 /2
and the ensemble average is evaluated over n, n, and Cn.
The Gaussian RVs Tc and Ts are uncorrelated and hence independent. Their common variance is E02 /2. The envelope of Ez
is given by
Tc2 + Ts2
(19)
f f 2 df
(23)
fm
|Ez | = V =
e j2 ( f f c )t
6
5
4
3
2
1
f () = f c + f m cos
(20)
0
f f
c
fm
212
RTT ( ) = Tc (t)Tc (t + )
+ Ts (t)Ts (t + )
fc + fm
3b
1
s
=
2 cos(2 ( f f c )t) df
f c f m 2 f m
f fc
1
fm
(24)
From Ref. 20, we have the result
1
1 x2
(25)
NV =
Vref
2
vp(V,
v)
dv = 2 f m e
(26)
Time (s)
Figure 10. A method to compute the level crossing rate and fade
duration for Rayleigh fading.
1
T i i
(27)
where v is the rate of change (the time derivative) of the envelope V(t), p(V, v) is the joint pdf of v and v at a specified value
of V, and V/Vrms is the normalized value of the envelope.
It can be verified that the peak value of NV occurs at
3 dB.
P [v V ] =
p(v) dv = 1 e
(28)
T fade =
2
1
1 [e 1]
i =
TNV i
2 f m
(29)
(30)
where Tc(t) is the in-phase component and Ts(t) is the quadrature component. The effect of the multipath fading channel is
represented by T(t), which is a complex-valued, multiplicative
scaling of the transmitted signal. The time-varying nature of
Tc(t) and Ts(t) manifests itself as a random phase modulation
of (t). In FM receivers, the effect of T(t) produces a baseband
noise component, which can be evaluated (1). The baseband
noise is characterized by the frequency deviation it produces.
It is interesting to note that the frequency deviation (and
hence, the rms value of the baseband noise) depends on the
depth of the fade. As the signal envelope V experiences deep
fades, the frequency deviation of the random FM increases
proportionally. The effect of T(t) can also be viewed as random
phase rotation given by arctan [Ts(t)/Tc(t)]. Hence, multipath
fading causes impairments of the received signal that have
the characteristics of random FM and random phase shifts.
In the preceding sections, the power spectrum and the envelope correlation of a Rayleigh faded signal have been discussed. Using these results, two methods are obtained for
generating Rayleigh fading:
f n = f m cos
2n
N
, n = 1, 2, . . ., N0
(31)
and one oscillator with frequency f m, where f m is the maximum Doppler frequency. The total number of oscillators
is (N0 1), where N0 (N/2 1). The parameter N is
chosen such that Tc and Ts are Gaussian and the power
spectrum of T(t) Tc(t) jTs(t) closely approximates the
condition in Eq. (22). In Ref. (1), it is shown that a good
approximation is obtained for N 34 ( N0 8).
In Fig. 12, the phases and n, n 1, 2, . . ., N0 are
chosen such that the probability distribution of the resultant phase (arg[T(t)]) is close to a uniform distribution in
[0, 2]. These phases are introduced by the respective
multipliers 2 cos n and 2 sin n. The uniform distribution of arg[T(t)] is ensured if
Tc2
= Ts2
and
Tc Ts
= 0
(32)
=0
The method for implementing the stochastic filtered noise approach is shown in Fig. 11. The steps are as follows:
n
N0 + 1
1
Scale1 =
2N0
1
Scale2 =
2(N0 + 1)
n =
Gaussian
noise source
(real)
213
Baseband
doppler filter
Tc
In-phase component
Tc + jTs
Gaussian
noise source
(real)
Baseband
doppler filter
Ts
Quadrature component
Figure 11. Stochastic filtered noise method for generating Rayleigh fading.
214
cos(2 f1t + 1)
2 sin 1
...
2 cos 1
cos(2 fNo t + No 1)
2 cos No
2 sin No
1
cos(2 fnt)
2
2 sin
2 cos
...
...
+
Scale2
Scale1
Tc(t)
Ts(t)
f No / f m
f1 / fm
1.0
1/ 2
...
1.0
fn / fm
Oscillator frequencies
c (1 , 2 ; t) =
1
E[c(1 ; t)c (2 ; t + t)]
2
(33)
(34)
=
c (k )
(35)
2 =
c (k )
(36)
(37)
1
1
2
6
1
BD
(39)
9
16 f m
(40)
215
(38)
216
those samples below the threshold. This model can be implemented by using a tapped-delay line or FIR model, thereby
allowing us to model any arbitrary channel.
Time flat
Time selective
Frequency selective
Frequency selective
LARGE-SCALE EFFECTS
Bc
Frequency flat
Time selective
Time flat
Frequency flat
Tc
Symbol duration (Ts)
Figure 13. A typical power delay profile and the method of sampling
the power delay profile to generate a tapped-delay line model.
Time delay
;
;;;
X (dB)
Understanding and characterizing the effects of the RF propagative channel are essential to designing RF communication
systems. A wide range of channel conditions are encountered
in RF communications, all the way from LOS channels to severely obstructed channels. Further, the channel may also be
time-varying. Hence modeling is based on statistical and experimental information. This is an area of extensive research
and measurements, over the past two decades, and even until
the present time (1,7,10,1620). In this section, the two main
components of signal variability due to the large-scale effects
of RF propagation, namely, path loss and shadowing, are discussed.
Scattering
;;
;;
Reflection
Diffraction
Receiver
Transmitter
Figure 14. Classification of multipath fading channels.
ing the propagative path loss (because the criteria for freespace propagation are not met). This topic has been very extensively studied. Detailed information can be found in Refs.
7, 10, 12, and 16. Computation of path loss is of particular
interest to communication systems designers. Because the actual RF communication environments encountered in practice
are so numerous, a unified theoretical/analytical framework
for estimating path loss is not feasible. Most system designers
resort to empirical approaches and semianalytical methods,
which have been validated by experimental/measured data,
to estimate the path loss. The work of Okamura and Hata
(13) is very widely used for path loss estimation. Okamuras
work is based purely on measured data, and Hata provided
the empirical model to fit that data. The advantage of using
empirical models and curve fitting to measured data is that
it accounts for both known and unknown sources of path loss.
On the other hand, the disadvantage is that the validity of
the empirical model, derived from a set of data, is not guaranteed for a different environment.
Let d be the distance between the transmitter and receiver. Both theoretical and measurement-based models show
that the average path loss [Lp(d)] increases directly proportional to dn, where n is called the path loss exponent. Typically n 2, as summarized in the Table 1. By contrast, in
free space, N 2. The path loss Lp(d) is given by
Pt Gt Gr 2
Pr (d) =
(4 )2 L
1
d2
(41)
(4 )2 L
Pt
= 10 log10
+ 20 log d
Pr
Pt Gt Gr 2
(42)
d 2
0
217
Lp (d) d n
Lp (d) = L p (d0 ) + 10n log10
(44)
d
(45)
d0
The reference point d0 is chosen such that Lp(d0) can be computed using the free-space path loss model.
Hata and COST-231 Models. This is one of the most widely
used models for estimating path loss in RF communication
channels. Based on extensive measured data, Okamura generated sets of curves that characterize the median attenuation (50th percentile) Lp,50, for a wide range of environments
(range of carrier frequency ( fc), effective height of transmitting antenna (ht,eff ), and distance d from transmitter. Hata
(12) provided an empirical formulation from Okamuras data,
which shows good agreement (between the model and the
measured data) for f c 1.5 GHz. An extension of Hatas
model for frequencies up to 2 GHz is provided in Ref. 14. The
Hata model and COST-231 models are given below: Hata
Model
(46)
, d d0
(43)
Outdoor Propagative Loss. In dealing with non-LOS environments, which is typical of most RF communication links,
such as cellular/PCS, we need appropriate models for comput-
Environment
Free space
Urban cellular
In-building (non-LOS)
Shadowed urban cellular
218
Pt
COST-231 Model
(47)
d
Lp (d) = Lp (d0 ) + 10n log10
+ (dB)
d0
Transmit power
Log-normal shadowing
Pmin
Receiver sensitivity
Thermal noise floor
Shadowing
As discussed earlier, shadowing is caused by terrain and
other environmental factors, such as foliage. The effect of
shadowing causes the variations in the mean of the received
signal. Let L p(d) denote the path loss (including the effect of
shadowing) at a specified distance from the transmitter.
Based on extensive measurements, it has been verified that
L p(d) can be characterized as a random variable with a lognormal distribution about the mean value. When expressed
in dB, the RV perturbing the local mean value of the path
loss is a normal RV, as given by
p (d) = [Lp (d) + ](dB)
L
(49)
(48)
Hata
COST-231
Carrier frequency f c
Effective transmit height h t,eff
Effective receive height h r,eff
Distance d from transmitter
Correction factors
150-1500 MHz
30-200 m
1-10 m
1 km
a(h r,eff )
1500-2000 MHz
30-200 m
1-10 m
1-20 km
a(h r,eff ), C M
f ( ) =
(50)
The path loss Lp is deterministic (based on the distance between the transmitter and receiver), whereas the fading and
shadowing are probabilistic. The amount of margin must be
judiciously chosen so that the net margin is minimized but
still meets the minimum signal strength requirements. If the
amount of losses exceeds the margin of a communication system, then an outage occurs, which implies that the communication link cannot be used until the channel conditions improve. This summarizes the tradeoffs that must be handled
by the designers of communication systems.
CONCLUSION
A comprehensive overview of the RF signal variations related
to propagation in multipath fading channels has been presented. The diverse phenomena that cause signal variations
are described via mathematical models. The different types of
fading and their salient features are discussed in detail. This
article provides a mathematical and an engineering-oriented
treatment of multipath fading, thereby providing the reader
with the necessary tools and the information to understand
the different RF propagative issues and the way they affect
wireless communication.
APPENDIX
Derivation of Rayleigh pdf
Let X and Y be two independent identically distributed (iid)
zero-mean Gaussian RVs with variance 2. Because X and Y
are independent, their joint pdf is the product of their marginal pdf s:
FX ,Y (x, y) =
2
2
1
x +y
2 2
e
2 2
(51)
X2 Y2 and
Using a change of variables, V
tan1
(Y/X), the joint pdf f V,(v, ) is given by
fV, (v, ) =
2
f X ,Y (x, y)
v
v
=
e 2 2
2
|det J (x, y)|
2
(52)
fV, (v, ) dv =
v e 2v 2
2
fV (v) =
1
2
v0
219
(53)
(54)
v<0
R. D. KOILPILLAI
Ericsson, Inc.
270
RECEIVING ANTENNAS
z
I(s')
RECEIVING ANTENNAS
r(s')
s1
R
s'
y
x
s2
Figure 2. Coordinate system for radiation field.
(a)
y
Aap
Einc
Z
x
ZL
E = j 30k
e jkR
I h
R 0
(1)
(b)
Figure 1. Reception of an electromagnetic plane wave. (a) Halfwavelength dipole antenna. (b) Array antenna on a circular cavity.
h=
1
+ (S )
}
{(S )
I0
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
(2)
RECEIVING ANTENNAS
with
S=
s2
)R
ds
(3)
s1
V02 = E1 z
= ( j 30 k
e jkR
I h) hd
R 01
(4)
Using Eq. (1), the radiation field from antenna 2 with terminal current I02 is written as
E2 = j 30 k
e jkR
I h
R 02 d
(5)
The open-circuit voltage V01 at antenna 1 induced by the radiation field from antenna 2 satisfies the relationship according
to the [reciprocity theorem (1)]
V01 I01 = V02 I02
OPEN-CIRCUIT VOLTAGE
271
(6)
e jkR
h I02 hd
R
(7)
Using Eq. (5) and replacing V01 and E2 with V0 and Einc, respectively, Eq. (7) is written as
V0 = Einc h
(8)
Ant. 1, h
V02
Ant. 2, hd
(a)
V01
I02
E2
Ant. 1, h
Ant. 2, hd
(b)
Figure 3. Determination of open-circuit voltage at antenna 1. (a) Antenna 1 for transmission. Antenna 2 is an infinitesimal dipole for reception. (b) Antenna 1 for reception. Antenna 2 is an infinitesimal
dipole antenna for transmission.
(9)
where Vrec ZLI and Vtrans ZAI, with ZA defined as the antenna input impedance. From Eq. (9), the terminal current I
272
RECEIVING ANTENNAS
Einc
Vrec
ZL
Einc
(a)
+
V0
WR =
Vtrans
(b)
(c)
Figure 4. A receiving antenna. (a) Original receiving antenna problem. (b) Antenna with open terminals. (c) Transmitting antenna.
is given as
I=
V0
ZA + ZL
(10)
(11)
ZA = (RA + rA ) + jXA
(12)
where RA and rA are the radiation resistance and the loss resistance (not contributing to the radiation), respectively.
Then, the power delivered to the antenna load is given as
|V0 |2 RL
WL =
|ZA + ZL |2
(13)
|V0 |2 (RA + rA )
|ZA + ZL |2
(14)
ZA
I
V0 = Einc h
(RA + rA ) + jXA
RL + jXL
(15)
|E h|2
|Einc h|2
= inc
4(RA + rA )
4 RL
(16)
ZL
|Einc h|2 RL Z0
|ZA + ZL |2 |Einc |2
|Einc h|2
Z0
Mimp
4(RA + rA )
|Einc |2
|E h|2
Z0
= inc
Mimp
4RA
|Einc|2
(18)
2
|Einc h|2
Ga
M
4
|Einc |2 |h|2 imp
(19)
(17)
Ar =
|Einc h|2
WL max
4 RL
WL =
2
Ga MpolMimp
4
(20)
The maximum receiving cross-section, which is called the effective area Aeff , is obtained when Mpol Mimp 1.
RECEIVING ANTENNAS
Aeff =
2
Ga
4
(21)
Aeff
1
Aap
273
G(, ) =
(a4)
Ga =
(a5)
(a6)
4|h(, )|2
|h(, )|2 d
(a7)
where
(22)
D=
Example. A parabolic reflector antenna has an aperture efficiency of less than 70% (68). Flat antennas, such as curl
and helical element array antennas (9,10), have aperture efficiencies ranging from 70 to 90%, where the elements are
arrayed using a circular cavity, as shown in Fig. 1(b).
APPENDIX
The input power Win[ I(0) (RA rA)] to a test transmitting
antenna is transformed to the radiation power Wrad( RA
I(0)2) with a relationship of Wrad Win, where is called the
radiation efficiency:
2
Wrad
RA
=
Win
RA + r A
(a1)
|E(R, , )|2
d
Z0
(a2)
k
4
2
Z0
|h|2 d
(a3)
HISAMATSU NAKANO
Hosei University
274
RECORDERS
360
REFLECTOR ANTENNAS
REFLECTOR ANTENNAS
Reflector antennas have been of importance for decades in
several areas of electrical engineering, ranging from telecommunications and radars to deep-space exploration and radio
astronomy. This is due to the high gain of reflector antennas,
typically above 30 dBi. If we extend the concept of a reflector
antenna to a reflecting mirror and view the human eye as the
feed antenna operating in receiving mode, reflector antennas
have been known for centuries. Optical astronomers have
long been using reflecting mirrors in telescopes to enhance
the visibility of stars, planets, and other celestial bodies.
The basic principle of operation of a parabolic reflector is
that all rays emanating radially from a point source located
at the focal point are reflected as a concentrated bundle of
parallel rays, which can propagate for very long distances
without loss due to speading. Inversely, incident rays parallel
to the axis of symmetry of the paraboloid are all reflected toward its focal point, which concentrates the received signal at
a single point. In that case, if the human eye or camera is
placed a little bit behind the reflector focal point, an image
with enhanced luminosity and definition is formed (Fig. 1).
However, reflector antennas can be designed to be wideband devices, not limited to operation at frequencies covered
by the spectrum of visible light. Radio telescopes, for example,
search for celestial radio sources over a wide range of frequencies (e.g., 300 MHz to 40 GHz). In this case, the radio sources
and corresponding frequencies are marked on charts ac-
Image
Figure 1. Basic principle of operation of a parabolic reflecting mirror. The paraboloid surface is formed by rotating the parabolic curve
about its axis of symmetry (s axis).
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
REFLECTOR ANTENNAS
(a)
(b)
S
(c)
S
(d)
Figure 2. The evolution of reflector antenna systems: (a) single axisymmetric reflector, (b) dual axisymmetric reflector, (c) single offset
reflector, and (d) dual offset reflector. The main reflectors are parabolic.
361
362
REFLECTOR ANTENNAS
Figure 3. Full-scale model of one of the twin Voyager spacecraft. Note at the center the dual-shaped, high-gain reflector
antenna employing a main reflector with a diameter of 3.66
m (12 ft). (Courtesy Jet Propulsion Laboratory. Copyright
California Institute of Technology, Pasadena, CA. All rights
reserved. Based on government-sponsored research under
contract NAS7-1260.)
Dp
Point F
Focal point
Point A
Point B
Point C
Point P
U L
U
Zf
D/2
C
B
Dp/2
D/2
H
L
L
Xf
S
F
Figure 4. Geometry for the axisymmetric (H 0) and offset parabolic reflector. See Table 1 for definitions of parameters.
Definition
REFLECTOR ANTENNAS
2F
= F sec2 f
1 + cos f
2
(1)
or
F rf cos2
f
=0
2
(2)
(3)
1
4F/Dp
(4)
L = 2 tan1
4H Dp
4F
n = r f cos f + f sin f
2
2
U = 2 tan1
1
4F/Dp
(5)
L + U
2
H
C = 2 tan1
2F
B =
f
2
(6)
cos r = cos
f
= z n
2
(7)
(8)
(9)
(10)
(11)
(12)
with
Dp = D + 2H
363
(13)
REFLECTOR ANTENNAS
(14)
(15)
Axisymmetric
Offset
Shape:
Projected diameter D:
Parent reflector diameter Dp :
F/Dp :
Offset of reflector center, H:
Parabolic
171
171
0.3
0
Parabolic
85.5
171
0.3
42.75
Linear (xf)
Gaussian; Eqs.
(14) and (15)
14.04
70
0
Linear (xf)
Gaussian; Eqs.
(14) and (15)
14.04
70
39.81
0.02
70
0.01
0.00
70
3
90
90
70
70
6
90
0.01
10
e jkr f
C(f )[f cos f f sin f ]
rf
XPOL
Ef =
COPOL
0.03
66
v = sin sin
364
20
0.02
40
0.03
0.03
0.01
0.01
u = cos sin
0.03
Figure 5. Contour plots in decibels of the computed co- and crosspolarized patterns of the 171 diameter axisymmetric parabolic reflector specified in Table 2.
REFLECTOR ANTENNAS
50
COPOL
XPOL
50
0.02
0.00
3
2
v = sin sin
2
0.01
x
50
3 6
35
2
0.01
S = sin1
10
35
0.02
20
50
50
0.03
0.03
0.01
0.01
u = cos sin
0.03
Figure 6. Contour plots in decibels of the computed co- and crosspolarized patterns of the 85.5 diameter just fully offset parabolic
reflector specified in Table 2.
symmetry is still present about the xz plane, and no substantial XPOL occurs in that plane. These results are demonstrated using the GRASP code for the offset reflector example
with the XPOL contour plot shown in Fig. 6, for which the
feed has a pointing angle of f B 39.81, computed according to Eq. (11). The feed again has the pattern given by
Eqs. (14) and (15), with a 10 dB beamwidth of 70. The computed peak XPOL is 22.4 dB relative to the copolarized
beam maximum of 47.39 dBi. Figure 6 indicates that the copolarized pattern is still symmetric, and the XPOL peaks are
located at the copolar 6 dB contour line.
This example is typical of single offset reflectors and shows
that single offset paraboloids illuminated by conventional
feeds are limited by XPOL performance (9). It is worth noting
that cross polarization arises from the reflector curvature and
from the tilting of the feed. A planar reflector, for instance,
does not depolarize an incident field coming in a direction perpendicular to the reflector. Thus we see that the XPOL decreases as the reflector curvature rate, F/Dp, increases. However, this reduction is not significant in offset reflectors, on
account of the substantial feed tilting normally encountered
in practice (15).
A XPOL level above 22 dB is often unacceptably high
(3,7). In the section titled Conditions for Minimizing Cross
Polarization in Offset Cassegrain and Gregorian Systems we
discuss procedures to reduce XPOL in offset parabolic reflectors. Next we discuss an important property of parabolic reflector antennas that is inherently related to XPOL.
Beam Squint. As we have seen, offset reflectors offer significantly reduced aperture blockage but introduce high
XPOL when illuminated by a LP feed. On the other hand,
offset parabolic reflectors fed by a circularly polarized (CP)
feed presenting a balanced radiation pattern do not have sub-
sin
f
(16)
2Fk
RHCP beam
LHCP beam
0
Relative pattern (dB)
0.03
365
10
0.7
20
30
40
10
2
2
Angle (deg)
10
366
REFLECTOR ANTENNAS
f
F
(17)
where BDF is the beam deviation factor, which can be approximately determined for small feed displacements f , in axisymmetric and offset paraboloids, by (3,20)
1 + 0.36 4
BDF =
1+ 4
F
Dp 2H
F
Dp 2H
2
2
(18)
S
Z
Figure 8. Beam squint generation mechanism.
Design of reflector antennas presents a challenge to the antenna engineer, especially in that so many parameters are
available for adjustment. The main purpose of this subsection
is to present a complete procedure to design axisymmetric
and offset reflectors, as well as to provide some insights into
the basic tradeoffs inherent in the process.
REFLECTOR ANTENNAS
367
Location
On Focus
Feed Type
Balance
Unbalanced
Off Focusa
Balanced
Unbalanced
Offset ( f 0)
On Focus
Balanced
Unbalanced
Off Focusa
Balanced
Unbalanced
Cross
Polarization
Beam
Squint
Linear
Circular
Linear
Circular
Linear
Circular
Linear
Circular
No
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
Yes
No
Yes
Linear
Circular
Linear
Circular
Linear
Circular
Linear
Circular
Yes
Nob
Yes
Yes
Yes
Yes
Yes
Yes
Nob
Yes
Nob
Yes
Nob
Yes
Nob
Yes
Beam deviation also occurs; see the subsection entitled Beam Deviation.
Except for small reflector antennas (i.e., D 12 and F/Dp 0.25); see Ref. 21 for further details.
15
49
48
20
47
46
25
45
44
30
43
Gain (dBi)
Polarization
42
35
Gain
XPOL
SLL
40
41
40
39
30
40
50
60
Feed pointing angle f (deg)
70
368
REFLECTOR ANTENNAS
decibels at a direction f as
4Ap
=
ap
2
D 2
(19)
f + f
+ 20 log cosq f
2
(20)
(23)
A usual value for RI at the reflector edges, often referred to as the edge illumination EI, is 11 dB and
assures optimal gain performance for axisymmetric paraboloids (f 0 and f L U); see Ref. 20. In
offset reflectors, L U, but an specified value of EI at
both edges can still be obtained by solving for f the
equation formed by imposing that q(f L) q(f
U). Once f is determined, the parameter q can be calculated directly from Eq. (23) for the specified value of
RI (i.e., EI), at either f L or f U, since they now
should yield the same result. Under this condition, the
edge illumination is called balanced and yields nearminimum sidelobe levels over practical ranges of feed
pointing, with only small penalties in gain and XPOL
(15). A graphical technique to determine f for the same
condition was introduced in Ref. 15 and is especially indicated when only measured feed patterns are available. In practice, however, it is common to find offset
systems employing f B, Eq. (11), or f C, Eq. (12).
For either case the edge illumination is in general unbalanced. As a consequence, different values of q are
obtained with Eq. (23), depending on whether f L
or f U. A simple arithmetic mean can then be taken
to specify the required feed pattern. Although approximate, this simple procedure yields reasonably good results in practice and is well suited for our purposes.
Once the parameter q is determined, Eqs. (22) and (19)
can be used to estimate the aperture diameter, or area,
of the required feed antenna. A typical value of ap
0.55 (55%) can be used for feed horns. A more exact
approach is available when the feed antenna is an openended rectangular waveguide of wide and narrow dimensions a and b, or an open-ended circular waveguide
of radius a. For those cases, the waveguide dimensions
can be determined from Eq. (22) with gf 32ab/2
(ap 0.81) or gf 10.5a2 / 2 (ap 0.84). If the result
indicates a feed antenna with an aperture considered
too large, a higher value of RI (i.e., EI at the reflector
edges) should be employed to avoid unnecessary
blockage. For an offset reflector, a larger value of H can
also be tried and the whole procedure needs to be repeated.
(21)
RI
+ f
log cos2 f
2
q = 20
log cos f
(22)
The aforementioned design procedure was successfully employed to obtain the preliminary design of a 1.6 m just fully
offset paraboloid, built and tested at the University of Brasilia for satellite TV reception at C band (Fig. 10). Nevertheless, the use of a suitable computer code before the manufacturing process is highly recommended to confirm the electrical
performance of the reflector system. Techniques often implemented in numerical codes for the analysis of reflector antennas are discussed in the section entitled Analysis Methods
and Evaluation.
As a final note, we mention that surface distortions from
ideal parabolic shapes are normally introduced in any manufacturing process. Random reflector surface errors can be al-
REFLECTOR ANTENNAS
Figure 10. Just fully offset parabolic reflector antenna for satellite
TV reception at C band, built and tested at the University of Brasilia.
(Courtesy of Carlos Muller, University of Brasilia.)
rs
D 2
(24)
rs = e
where
(2 s ) 2
369
(25)
The parameter S is the root mean square (rms) surface deviation and is approximately one-third of the peak-to-peak error
(20). Detailed information on other efficiency factors left in
ap can be found in the literature (20). Mass production of reflectors using presses and molds to shape glass fibers and
other components normally yields S 0.01 (rs 0.98 or
98%). Errors larger than that can cause significant gain loss
and pattern deterioration due to the introduction of phase errors. Machined metal reflectors are very accurate and have
S 0.04 mm. Reflector antennas are wideband devices by
nature and are limited at the upper frequencies by the
smoothness of the reflector surface and at the lower ones by
the reflector electrical size. The operating bandwidth of a reflector antenna in practice is normally set by the bandwidth
of the feed antenna employed to illuminate the reflector. Although herein illustrated for parabolic reflectors, random surface errors occur in any type of reflector antenna, such as the
ones discussed next.
Other Single-Reflector Systems
A few other types of parabolic reflectors can also be obtained
from Fig. 4, including offset configurations. The parabolic cyl-
370
REFLECTOR ANTENNAS
(26)
D/2
Dp/2
z
ZS
fS
D/2
Xf
2c
A
AS
F1
F2
XS
DS
B
Zf
REFLECTOR ANTENNAS
taneously minimizes XPOL and spillover loss (i.e., feed radiation missing the subreflector):
Definition
Diameter of the projected aperture of the parabolic main
reflector
Dp
Point F1
Point F2
Point A
Point AS
Point B
Point C
U L
DS
FS
Angle between the main reflector and feed axes (s and zf)
|e2 1| sin
(1 + e2 ) cos 2e
(27)
where e is the subreflector eccentricity (0 e 1 for an ellipsoid and e 1 for a hyperboloid), e c/( fS c). Equation (27)
is generally referred to as the Mizugutch condition and has
the following alternative form, known as the Dragone condition (24,26):
tan
e+1
=
tan
2
|e 1|
2
tan
=
2
e 1 2
e+1
tan
+ C
2
(28)
(29)
371
(30)
372
REFLECTOR ANTENNAS
Figure 13. Artwork of the Green Bank Radio Telescope reflector antenna. The dual offset Gregorian configuration employs an offset parabolic main reflector with a 100 m projected aperture
diameter. (Courtesy of George Behrens, National Radio Astronomy Observatory.)
REFLECTOR ANTENNAS
Offset parabolic
100
208
60
54
PRAC
GRASP
373
tions and approximations inherent in each of these formulations, as well as on their intrinsic differences.
The GO technique yields the aperture fields, assuming
equal angles of incidence and reflection. The far-field patterns
can then be calculated using a Fourier transformation directly, which is equivalent to obtaining equivalent currents
and then integrating, as described later in this subsection.
With the use of image theory, it is necessary to know only the
electric field distribution over the reflector projected aperture, Er, which is computed from the incident electric field Ei
(i.e., the feed radiation), with (20)
Linear (xf)
cos4.58; Eqs.
(14) and (21)
13.08
77.92
42.77
Gain Gf (dBi):
10 dB beamwidth (deg):
Feed angle f (deg):
Linear (xf)
Gaussian; Eqs.
(14) and (15)
13.14
77.92
42.77
System Performance
Gain G (dBi)
Cross-polarization level,
XPOL (dB):
Sidelobe level, SLL (dB)
Aperture efficiency ap , %
82.87
21.54
82.79
21.56
26.72
78.48
27.21
77.05
E r = 2(n E i )n E i
where n
is the unit vector normal to the surface; see Eq. (5).
Equation (31) assumes that at the point of reflection the reflector is planar and perfectly conducting. In addition, the incident wave from the feed antenna is treated locally as a
plane wave. These same assumptions are also used by the PO
technique to determine the surface currents, Js, over the reflector as follows:
J s = 2n H i
(32)
(31)
374
REFLECTOR ANTENNAS
E (rr ) = j
e jkR
(I rr )
2 R
(33)
s
REFLECTOR ANTENNAS
375
Figure 14. Just fully offset parabolic reflector antenna with a projected aperture diameter of 1.8 m. (Courtesy of Nick Moldovan, Prodelin Corporation.)
COPOL (PRAC)
XPOL (PRAC)
COPOL (meas.)
XPOL (meas.)
Ef =
10
Relative pattern (dB)
where CE(f ) and CH(f ) denote the feed-pattern cuts in the Eand H-planes, respectively. A yf -polarized feed pattern, as
well as circularly polarized ones, can be obtained from Eq.
(34) by introducing the modifications already suggested in the
subsection entitled Cross Polarization. Note that Eq. (34)
reduces to Eq. (14) for CE(f ) = CH(f ) C(f ).
Finally, we can approximate even further the electrical behavior of feed antennas, although still ideally modeled with a
fixed phase center, using the complex polarization ratio pr,
defined as
20
30
40
(34)
0
Angle (deg)
(35)
(36)
376
REFLECTOR ANTENNAS
jkr f
(37)
rf
and
(38)
where ECO and ECROSS are the co- and cross-polarized radiation
patterns, respectively. It is worth mentioning that Eq. (37)
reduces to Eq. (14) for CE(f ) CH(f ) C(f ). In addition, Eq.
(38) yields the cross-polarized peaks in the 45 planes, which
is consistent with measured data and more sophisticated theoretical models (33). Nevertheless, it is not our intention to
analyze feed antennas completely, but rather to present simple analytical models that can be useful to antenna engineers
as a first approximation to real feed patterns.
As discussed in the subsection titled Summary of Main
Results, balanced feeds yield improved performance when illuminating reflector systems. A type of balanced feed antennas widely used in practice is the corrugated conical horn,
also often referred to as a scalar horn (Fig. 16). Corrugated
conical horns present a phase center that is reasonably stable
with changing frequency (23,33), in addition to a copolarized
pattern that is nearly balanced for practical purposes and can
be well modeled by Eqs. (14) and (15) or (21).
The main purpose of using corrugations is to obtain the
same boundary conditions around the inside of the horn. For
corrugation depths of a quarter wavelength, the short circuit
at the bottom is transformed to an open circuit at the top of
the corrugation, yielding boundary conditions that appear to
be more uniform as the number of corrugations per wavelength increase (20). This yields a symmetric radiation pattern down to as low as 25 dB over a reasonably wide operational bandwidth, typically of 1.6 : 1 or more.
The cross-polarized pattern, however, may contain peaks
in the 45 planes, similarly to the unbalanced model of Eq.
(38). As discussed in the section entitled Multiple-Reflector
Antenna Systems, and as approximately modeled by Eq.
(30), feed XPOL peaks appear in the far-field patterns of dualreflector systems, even with the enforcement of the conditions
given by Eqs. (27) to (29). To minimize feed XPOL, a careful
design and construction process must be performed. The corrugation depths must be set properly to achieve resonance,
taking into account all dimensions related to the corrugations
as well as the general geometry of the horn (23). The process
has been successfully accomplished in practice, often also employing sections of tapered and/or dual-depth corrugations,
yielding horns presenting balanced copolar patterns and
XPOL levels below 35 dB over practical operational bandwidths (23,34).
Finally, other types of feed antennas are also used in practice, such as wire antennas and pyramidal and sectoral horns
(20). The latter are used for illuminating reflectors with elliptical projected apertures, which were discussed previously. In
addition, it is common to have an array of horns or other feed
antennas illuminating shaped reflectors or large reflectors,
such as radio telescopes. This is because the array yields better control of the phase distribution employed to illuminate
the reflector, enhancing beam contour and beam scanning
performance. A contoured beam is required to illuminate
properly a specified region of the earth, as seen from a satellite, and can be accomplished with the shaping of the reflector, one of the topics addressed in the following section.
ADVANCED TOPICS AND RESEARCH
Reflector Antenna Upgrading
During the past few decades reflector antenna designs have
evolved through several configurations to increase performance and/or reduce structural complexity. Electrical parameters that are of prime interest are aperture efficiency, SLL,
and, more recently, XPOL. All topics herein discussed apply
to the various types of reflectors previously addressed. However, the offset configuration is likely to retain, in the near
future, the largest percentage of the reflector antenna
market.
We first focus our attention on XPOL. Reflector antennas
presenting low XPOL (e.g., XPOL 35 dB) are necessary
for frequency-reuse applications, in which an overlap of orthogonally polarized channels is permitted. Many efforts are
being conducted to develop these kinds of antennas for mass
production (7,35). Dual offset reflectors can be designed for
low-cost construction, provided that specific manufacturing
constraints are carefully taken into account (7), an effort only
possible due to increased interest from industry. Single offset
reflector systems illuminated by a matched feed (23) or a feed
with a lens (35) can also be designed to satisfy stringent requirements on XPOL, yielding very compact designs. In the
latter case, the lens is designed to replace the subreflector,
and in both cases bandwidth performance is not as straightforwardly obtained as with the dual reflector configuration.
Research continues to be conducted within the area, yielding
innovative solutions that provide satisfactory XPOL performance while attending to practical manufacturing specifications. Cost-effective solutions normally require that attendance to a particular specification, such as low XPOL, be
achieved with minimal capital outlay, which implies using the
maximum amount of infrastructure and technology already
implemented. As mentioned previously, there is a tendency to
REFLECTOR ANTENNAS
employ existing molds for the main reflector, a concept referred to as reflector upgrading (7).
Within this context, existing single offset reflector molds
are normally used to construct the main reflector of a dual
configuration. However, many such molds are for just fully
offset geometries, which, in general, produce a dual reflector
configuration that is Gregorian with the feed axis zf intersecting the main reflector (Fig. 11). The same problem may
also occur in certain Gregorian configurations even when the
main reflector is not just fully offset. The final design should
provide suitable clearance between the bottom of the main
reflector and the feed axis in order to access the feed antenna
with a straight section of waveguide, thus reducing the complexity and cost of the manufacturing process. This setup is
achieved by rotating the parent ellipsoid (i.e., the conical surface from which the subreflector is generated) until the desired clearance is obtained. The rotation is performed in such
a way that the feed remains pointed toward the intersection
of the new subreflector and the ray coming from the center of
the main reflector, thus avoiding the introduction of spillover
and phase errors. The amount of rotation R that yields a desired angle between the main reflector and feed axes can
be determined from (7)
1+e
f
1 e cos(180 R C ) S
1+e
f
2(c + f S )
1 e cos(180 R C ) S
sin(180 R C ) = sin( + R + )
(39)
377
system. It is common nowadays to find feeds that already include low-noise amplifiers, downconverters, and other electrical devices in a single unit.
Shaped, Deployable, and Frequency-Selective Reflector Surfaces
Reflector antennas can be shaped to improve gain and aperture efficiency (e.g., Fig. 3). The basic concept is to shape the
subreflector of a dual-reflector system to obtain the desired
amplitude distribution at the main reflector projected aperture, normally nearly uniform for maximizing gain, and then
shape the main reflector to recover the uniform phase distribution, which was disturbed by shaping the subreflector. Resulting aperture efficiencies are usually in the range of 70%
to 80%, although higher values have been reported for dual
shaped offset systems (11). The fundamental tradeoff that antenna designers must face, in this case, is the achievement of
such efficiencies while still obtaining satisfactory sidelobe
levels.
Another topic certain to continue receiving attention in the
coming millenium is the shaping of single-reflector systems
for contoured beam applications. A contoured beam is necessary to illuminate specific regions of the earth efficiently from
a satellite. This avoids unnecessary coverage of regions outside the satellite main service area. Reflector shapes for contoured beam and high-gain applications, especially for the offset case, are normally obtained numerically through
elaborate synthesis and optimization processes (35,9,10),
which include the feed antenna or array. Equation (17) can
be used to set the initial positions for the feed antennas (3).
Also important is mathematical and numerical modeling of
the surfaces to yield results that can be implemented in practice (36). Feed arrays are also used with reconfigurable mesh
surfaces and deployable reflectors to enhance performance (1).
Reconfigurable shapes, deployable reflectors, and polarization- or frequency-selective surfaces require an in-depth study
of electrical and magnetic materials, which show that reflector antenna engineering is not possible without large interdisciplinary activities. Frequency-sensitive surfaces, for example, reflect only radiation in specific frequency bandwidths (1).
A four-frequency reflector system has been used for the ESA
NASA Cassini mission to Titan, the largest moon of Saturn
(1). The subreflector carried by the Voyager spacecrafts shown
in Fig. 3 is another example; it is transparent to radiation at
S band and reflecting at X band. The subreflector was manufactured with X-band resonant aluminum crosses etched on a
Mylar sheet, yielding reflecting and transmitting losses lower
than 0.1 dB.
Reflectors that can be deployed in space deserve distinct
attention. The classical solution is the so-called umbrella reflector (1), which works well for axisymmetric configurations.
An idea that seems promising for the offset case is a configuration that opens like a manual fan. In addition, inflatable
reflectors have been investigated over the past few years,
with the successful construction and testing of a few prototypes.
Neural Networks, Fuzzy Logic, and Genetic Algorithms
Applied to the Synthesis of Reflector Antennas
As we have seen, reflector antenna applications range from
very specific and unique systems, such as deployable reflectors and radiotelescopes (see the subsection on the Green
378
REFLECTOR ANTENNAS
Bank Radio Telescope), to large-scale production for commercial systems, such as earth terminals in VSAT networks and
small receiving antennas for satellite TV. In addition, reflector antennas are also used in radar systems and other devices
directly related to electronic warfare and defense. Due to the
wide range of applications and commercial importance, it is
essential to have reliable alternative algorithms to design effective reflector configurations. Analytical and numerical tools
that have recently been used with applied electromagnetics
and appear to work well for the synthesis of reflector antennas, although they are most familiar in other areas such as
controls and signal processing, include neural networks, fuzzy
logic, and genetic algorithms (3740).
The basic idea of using neural networks for designing reflector antennas is as follows: First one relates a few radiation
patterns, or other parameters of interest, directly to the reflector geometries that generated them according to a previously selected analysis algorithm. This process is normally
referred to as training the neural network. Once the training
is completed, the desired radiation pattern is employed as the
input parameter, so that the neural network yields the corresponding reflector geometry. Finally, the selected analysis
technique is used on the resulting geometry to validate the
process. This type of synthesis has been successfully implemented to determine reflector shapes for contoured beam applications (37). Reflector antenna synthesis based on neural
networks is normally very fast, because analysis techniques
are required only for the training of the network.
Also currently under investigation is the effectiveness of
using neural networks at a real-time level to reduce XPOL,
noise, and interference in single offset reflector systems.
Fuzzy logic can be applied to the synthesis and enhancement
of reflector antenna systems in a similar manner, although
there are certain inherent differences not addressed herein
(38).
Genetic algorithms (GAs), on the other hand, rely on an
optimization search to elect a suitable design. The initial set
of configurations, referred to as the basis population, is
formed by relating random designs to chromosomes, each one
formed by a sequence of binary numbers that define the corresponding reflector geometry. A figure of merit is then associated with each chromosome, becoming higher as the electrical
performance of the configuration comes closer to the desired
one. Chromosomes with the highest figures of merit are selected to cross with the remaining ones, a process called crossover (39,40). In addition, the best chromosomes (i.e., the ones
with the highest figures of merit) are often duplicated before
crossover, eliminating the worst ones, a procedure referred to
as natural selection (39,40). After the crossover is completed,
a few binary numbers in a subset of chromosomes can be randomly altered, simulating the process of mutation in biological evolution. All figures of merit are then recomputed, and
the whole process is repeated if the desired performance has
not been achieved.
The aforementioned procedure is a basic description of a
genetic algorithm, although variants are often encountered in
practice (39,40). A major advantage of GAs over other optimization techniques is that once the process has converged,
more than one configuration attending the desired performance is often encountered. It is also common to find configurations that are very different from classical solutions and
could not be easily conceived through straightforward reasoning. Practical manufacturing constraints can then be consid-
ACKNOWLEDGMENTS
The author is deeply indebted to George Behrens (National
Radio Astronomy Observatory) and Nick Moldovan (Prodelin
Corporation), as well as to their institutions, for providing
many illustrations and photographs reprinted in this work.
The author likewise thanks NASA/JPL and the Brazilian
Telecommunications Center for Research and Development
(CPqD/TELEBRAS).
BIBLIOGRAPHY
1. W. V. T. Rusch, The current state of the reflector antenna art
entering the 1990s, Proc. IEEE, 80: 113126, 1992.
2. S. Silver (ed.), Microwave Antenna Theory and Design, New York:
McGraw-Hill, 1949.
3. Y. Rahmat-Samii, Reflector antennas, in Y. T. Lo and S. W. Lee
(eds.), Antenna Handbook, New York: Van Nostrand Reinhold,
1988.
4. C. Scott, Modern Methods of Reflector Antenna Design, Norwood,
MA: Artech House, 1990.
5. P. S. Kildal, Synthesis of multireflector antennas by kinematic
and dynamic ray tracing, IEEE Trans. Antennas Propag., 38:
15871599, 1990.
6. K. W. Brown, Y. H. Lee, and A. Prata, Jr., A systematic design
procedure for classical offset dual reflector antennas with optimal
electrical performance, in IEEE Antennas Propag. Soc. Symp.
Dig., Ann Arbor, MI, 1993, pp. 772775.
7. M. A. B. Terada and W. L. Stutzman, Computer-aided design of
reflector antennas, Microw. J., 38 (8): 6473, 1995.
8. W. V. T. Rusch and P. D. Potter, Analysis of Reflector Antennas,
New York: Academic, 1970.
9. A. W. Love (ed.), Reflector Antennas, Piscataway, NJ: IEEE
Press, 1978.
10. B. S. Westcott, Shaped Reflector Antenna Design, London: Wiley, 1983.
11. A G. Cha, Preliminary announcement of an 85 percent efficient
reflector antenna, IEEE Trans. Antennas Propag., 31: 341342,
1983.
12. A. H. Rana, J. McCoskey, and W. Check, VSAT technology,
trends, and applications, Proc. IEEE, 78: 10871095, 1990.
13. A. C. Ludwig, The definition of cross polarization, IEEE Trans.
Antennas Propag., 21: 116119, 1973.
14. TICRA Eng., GRASP7Single and Dual Reflector Antenna Program Package, Copenhagen, Denmark.
15. M. A. B. Terada and W. L. Stutzman, Design of offset-parabolicreflector antennas for low cross-pol and low sidelobes, IEEE Antennas Propag. Mag., 35 (6): 4649, 1993.
16. A. W. Rudge and N. A. Adatia, Beam squint in circularly polarized offset parabolic reflector antennas, Electron. Lett., 11: 513
515, 1975.
17. D. W. Duan and Y. Rahmat-Samii, Beam squint determination
in conic-section reflector antennas with circularly polarized feeds,
IEEE Trans. Antennas Propag., 39: 612619, 1991.
18. W. L. Stutzman, Polarization of Electromagnetic Systems, Norwell, MA: Artech House, 1993.
MARCO A. B. TERADA
University of Brasilia
379
666
SATELLITE ANTENNAS
SATELLITE ANTENNAS
Antennas for satellite systems have developed and increased
in performance over the last 40 years and greatly contribute
to satellite capabilities. The preponderance of this antenna
development has been for communication systems, and other
development has been for remote sensing applications. Early
satellite systems provided not only proof of concept but also a
useful initial operating capability. The payload constraints of
early satellites together with primitive attitude control systems limited antennas to simple narrow-bandwidth, broadcoverage designs, a far cry from todays sophisticated antenna
systems with substantial throughput capabilities. Similarly,
early users needed large ground terminal antennas to achieve
the required link performance, and terrestrial links were provided to other system users. Today, systems are being developed to communicate to individual users with hand-held terminals (1).
Both spaceborne and user antenna systems have requirements and development issues that are unique to satellite
systems. In addition, satellite system antennas have testing
requirements and objectives that differ from other antenna
applications. Antenna technology, requirements and development issues, and test methods and objectives are described.
SPACE SEGMENT ANTENNAS
Spaceborne antennas have unique requirements and development issues. Like all satellite subsystems, much design atten-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
SATELLITE ANTENNAS
667
668
SATELLITE ANTENNAS
SATELLITE ANTENNAS
1000
800
600
669
20 Elevation
Separation (nmi)
400
Gain loss
200
100
80
60
Interference
source
40
Separation
Subsatellite
20
Desired
source
3 dB gain loss
10 dB gain loss
2
1 0.8 0.6 0.4
Antenna beamwidth (deg)
0.8
0 10
16
0.6
22
25dB
0.4
Latitude (deg)
0.2
0
U3
10
0
16
0.2
J7
22
0.4
0.6
0.8
0.8 0.6 0.4 0.2
0 0.2 0.4
Longitude (deg)
0.6 0.8
0.2
0
9
6
3
3
0
36
3
6
3
6
9dB
9
J7
0.2
0
3
0.4
6
0.6
6
3
0.4
Latitude (deg)
10
0
3
30
6
3
6
6
0.8
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8
Longitude (deg)
(b) Thinned array contours
Figure 2. Comparison of multiple beam antennas and thinned array
performance (after Ref. 8).
SATELLITE ANTENNAS
103
670
102
3 dB
crossover
10
5 dB
crossover
Earth
coverage
1
0.5
10
20
that active tracking of a ground beacon is required to compensate for satellite attitude variations.
The ability to process and direct the information on the
collection of multiple beams requires beamforming networks
that separate information on the uplink beams and combine
information on the downlink beams. A wide variety of beamforming networks has been developed. The simplest design is
a fixed beam-former network that simply combines beams in
a fixed pattern. This beam former together with transponder
channelization maps the uplink beam patterns into the downlink patterns. other beam-former designs use cascaded networks on switches or variable power dividers so that the beam
patterns can be changed on-orbit. Still other beam formers
use adaptive processing on the uplink beam former to negate
interference. Yet other systems route the uplink information
to a terminal on the ground, referred to as a gateway, where
the uplink beams are mapped into the downlink collection.
The beam formers have different requirements for the
uplink beams and downlink beams. The system noise temperature on the uplink can be established by preamplifiers which
are followed by the beam formers. In this way the system sensitivity is not degraded by the insertion loss of the beam formers, and lightweight networks and diode switching can be
used in the beam-former implementation. The downlink
beam-former network must be placed between the transmitters and the beam ports, so that power handling and insertion
loss are important design parameters. In addition, switching
for redundant transmitters is also required. These requirements depend on the application.
The development of multiple beam antenna systems has
been principally pursued for geosynchronous satellites, and
future design challenges exist to develop multiple beam designs for LEO and MEO satellites. While offset reflector an-
SATELLITE ANTENNAS
671
672
SATELLITE ANTENNAS
SATELLITE ANTENNAS
antenna beamwidth, the uncertainty in the satellites location, and the orbital dynamics. For communication applications, a pointing accuracy of one-tenth of a beamwidth is typically specified to minimize signal loss caused by pointing
errors.
In some cases, antenna pointing is trivial. For example,
the antennas for personal communication systems are purposely designed with broad coverage antennas so that the
user has no pointing requirements. Another common example
is a small user antenna for direct broadcast satellite reception
from geosynchronous satellites. These antennas are roughly
aligned using the users geographic location, verified by signal
peaking techniques, and rigidly secured. The combination of
the relatively wide beamwidth from the small antenna and
the stationary position of the geosynchronous satellite and its
station keeping results in a fixed pointing requirement.
In other applications where satellites have inclined orbits
or orbits that are not geosynchronous, the satellite undergoes
dynamic motion with respect to the user, and a means must
be provided to follow the satellite in orbit. For users with relatively broad beamwidths compared to the orbital uncertainty,
the knowledge of the satellite ephemeris and the user location
may be adequate to simply command the users antenna position. This open loop procedure is commonly referred to as program tracking.
If some uncertainty exists in the antenna pointing, such
as ephemeris data that are not current, another open loop
technique referred to as step track is commonly used to verify
correct alignment with the satellites position. The users antenna is pointed at the nominal position of the satellite as
might be obtained from program track. The antenna is then
commanded to move in equal and opposite angular positions
from this nominal pointing direction. If the antenna is correctly aligned with the satellite, the received signal level
should be reduced by the same amount at both angular offsets. If the signal levels at both positions are not identical,
the difference in the power level provides the angular correction to the nominal position. This process is repeated in the
orthogonal plane to properly position the antenna in two angular coordinates.
These two open loop antenna pointing techniques are commonly used together. The step track procedure is periodically
exercised to validate the correctness of the program track
pointing. A significant advantage of these techniques is that
minimal equipment is required. The antenna positioner is required and simple software commanding is needed to execute
the angular offsets. The power measurements at the different
angular offsets may be obtained from a simple measurement
of the receivers automatic gain control (AGC) voltage with
the appropriate calibration and linearity verification.
When user antennas have beam widths that approach ten
times the uncertainty in satellite position or a significant
amount of orbital dynamics exist, closed loop tracking techniques are required. These techniques are commonly used in
radar systems to locate targets and are referred to as monopulse designs. This term is derived from radar applications
because the tracking information is obtained from each radar
pulse. In radar applications, the received signal level varies
as the target changes aspect angle and typically has a significant dynamic range. Thus, tracking information is derived
from each radar pulse so that the dynamics of the target return do not degrade antenna pointing performance. Mono-
673
pulse systems for communication applications have easier requirements than those for radar systems. The received signal
level has a relatively constant amplitude, so that the monopulse signals can be sequentially sampled reducing hardware
requirements. The antenna pointing accuracy depends on
aligning the antenna with sufficient accuracy to minimize signal loss as compared to locating a radar target with the maximum precision practical.
Monopulse systems operate by forming two types of antenna beam shapes: a sum beam that receives the desired signal and a difference beam that has a null on the antenna
boresight axis. The signal power in the sum beam is maximized by positioning the antenna to the null of the difference
pattern. The ratio of the signal levels in the sum and difference beams is independent of the power density of the received signal, and provides a measure of the angular displacement from the antennas axis. This ratio can be used in a
closed loop system so that the antenna will align to the received signal and will follow any variations in the signal location.
Two different types of feed designs are commonly used to
generate the sum and difference beams. One feed design (27)
uses a conventional feed for the sum beam and an additional
four small feed elements to produce the difference beam. The
second design (28) uses a higher order waveguide mode to
produce the difference beam which is sampled by couplers. In
both cases, the difference beam outputs are coupled into the
sum channel and switched sequentially. The resulting AM
modulation on the sum signal is processed to derive the antenna pointing measurements.
674
SATELLITE ANTENNAS
These three test phases have a scope that greatly exceeds the
normal testing to define the component level performance of
antennas. This situation is particularly true with the trend
toward incorporating payload electronics into the antenna
system which results in the antenna system performance that
depends not only on the antenna components but also the
electronics performance.
The development testing has the objective to demonstrate
the designs fully comply with the system specifications. The
testing in this phase of the program is most closely related
to conventional antenna testing and quantifies gain values,
pattern characteristics, polarization purity, impedance properties, bandwidth variations, etc., as is typically performed.
These RF characteristics can generally be performed using
conventional RF test facilities and general purpose instrumentation. The antenna components are designed on a prototype or engineering model basis, so that weight projections
and compliance with thermal and mechanical requirements
for launch can also be evaluated.
The qualification testing has the objective of determining
whether or not the flight hardware is capable of the performance established for the design whose compliance has been
demonstrated in the development testing. An important part
of the program test planning is a requirements flowdown that
identifies the testing necessary to demonstrate overall system
compliance and assurance of flight worthiness. The principal
concern in this phase of the testing is how to perform the
testing without risk to flight hardware. Moving flight hardware to conventional antenna test facilities poses an unacceptable risk in many cases. Testing very lightweight antenna
designs may be precluded because they can be destroyed by
wind or become contaminated and the effects of gravity can
degrade the ability to project on-orbit performance. The development of portable near field test facilities is viewed as a need
to permit detailed antenna testing in payload assembly areas.
Another important part of the test planning in the early part
of the program is the identification of test ports for system
evaluation and the required development and calibration of
any specialized test fixture. Qualification testing must also
evaluate performance variations in the thermal variations
and vacuum conditions, and the ability to survive the shock
and vibration levels experienced during launch. Other issues
such as the reliability of deployment mechanisms if used
must be demonstrated.
On-orbit testing is conducted using specialized test terminals, which are often a part of the satellite control networks.
The testing at this phase is generally conducted shortly after
the satellite arrives at its orbital location. This testing generally concentrates on system level performance parameters,
such as uplink G/T, downlink ERP, and payload antenna
pointing alignment, to establish overall performance compliance. After this initial performance evaluation, these test
facilities are used to periodically assess on-orbit performance,
and together with the telemetry information provide a means
for on-orbit diagnostics. An important part of the test planning is an examination of the adquacy of the telemetry and
test terminal information for identifying failed components
and their potential substitution with redundant components.
Satellite systems also have some specialized testing requirements peculiar to their environment. One problem with
satellite systems is the static charging of components on-orbit. Typically, this charging occurs on dielectric materials
used for thermal protection not only for the antenna but also
other satellite subsystems. The charge builds up on these
components until discharge occurs. This phenomenon is referred to as ESD (electrostatic discharge), and the spectral
components of the discharge may have sufficient intensity to
interfere with payload receivers. The ESD spectral content is
typically measured by embedding a sample of the material in
an electron beam and using a pickup probe antenna and a
spectrum analyzer or very wide bandwidth oscilloscope.
Another problem which results from high power operation
in vacuum conditions is multipaction. In this case, the RF
energy may be sufficient to strip electrons from the surfaces
of materials exposed to the high power, which will damage
the component. The potential for multipaction is typically
evaluated by testing the components associated with high
power in an evacuated bell jar and using a spectrum analyzer
to observe any RF noise associated with multipaction. Such
testing can also evaluate any power handling and microarcing
limitations of the components by examining the components
after the test for damage. Temperature increases from highpower operation can also be measured.
A third problem is also associated with high-power operation and is referred to as PIM (passive intermodulation). This
problem results from exposing junctions with high power. Dissimilar metals or contamination forms weak diode junctions
whose nonlinearity generates intermodulation products. Such
junctions can occur in transmission line components, filters
and diplexers, and joints within the antenna structure. Those
components exposed to high power are tested by injecting two
tones at high-power levels (higher power transmitters than
the operational ones are sometimes used to insure the observed intermodulation does not result from products generated by a saturated transmitter) and observing any intermodulation products generated by the components under test
with a spectrum analyzer.
User antenna testing also has some unique issues for satellite systems (29). One problem results from testing on ground
terminal antennas, where the physical size of the antenna
precludes testing in conventional antenna test facilities. A
test technique using the emission from astronomical sources
is used to measure the G/T (30). The flux density of these
emissions has been well established, and a variety of sources
exist. The emission from the sun has a high level, but the
angular width of the solar disk limits this technique to relatively wide beam widths. Other sources such as the moon and
radio stars such as Cassiopeia A are used for narrower beam
widths. The positions of these sources are also well known,
and thus, the opportunity to validate the accuracy of the antenna positioning systems is also available. The antenna gain
in the transmit bandwidth can also be established by connecting a low noise preamplifier in place of the transmitter, measuring the G/T, separately measuring the system temperature, and multiplying to obtain the transmit antenna gain.
While this technique was developed for large ground terminal
antennas (31), with the available low noise preamplifiers, useful measurements can be made on much smaller antennas.
This technique is particularly useful for those antennas that
follow the trend of integrating the antenna feed and receiver
front end without connectors for test purposes.
Other electromagnetic measurements are required in the
development of both user and spaceborne systems to validate
EMI/EMC (electromagnetic interference/electromagnetic
675
SUMMARY
BIBLIOGRAPHY
1. F. Ananasso and F. D. Priscoli, The role of satellites in personal
communication services, IEEE Selected Areas Commun., SAC-13:
180195, 1995.
2. A. W. Love, Electromagnetic Horn Antennas, New York: IEEE
Press, 1976.
3. C. T. Brumbaugh et al., Shaped beam antenna for the global positioning satellite system, 1976 IEEE AP-S Symp. Digest, Amherst,
MA: pp. 117120, October 1115, 1976.
4. K. Ueno et al., Design and characteristics of a multiband communication satellite antenna system, IEEE Trans. Aerosp. Electron.
Syst., AES-31: 600607, 1995.
5. T. Katagi and Y. Takeichi, Shaped beam horn reflector antenna,
IEEE Trans. Antennas Propag., AP-23: 757763, 1982.
6. O. M. Bucci et al., Antenna pattern synthesis: A new general approach, Proc. IEEE, 82: 358371, 1994.
7. J. T. Mayhan, Area coverage adaptive nulling from geosynchronous satellites: Phased arrays versus multiple beam antenna,
IEEE Trans. Antennas Propag., AP-27: 410419, 1986.
8. K. M. SooHoo and R. B. Dybdal, Resolution performance of an
adaptive multiple beam antenna, IEEE MILCOM 89 Symp. Dig.,
Session 7, Boston, MA.
9. R. B. Dybdal, Multiple beam communication satellite antenna
systems, IEEE ICC Conf. Dig., 27D-1-5, Minneapolis, MN: June
1719, 1974.
10. A. W. Love, Reflector Antennas, New York: IEEE Press, 1978.
11. C. M. Rappaport, An offset bifocal reflector antenna design for
wide angle beam scanning, IEEE Trans. Antennas Propag., AP32: 11961204, 1984.
12. R. Jorgensen, P. Balling, and W. J. English, Dual offset reflector
multibeam antenna for international communications satellite
applications, IEEE Trans. Antennas Propag., AP-33: 1304
1312, 1985.
13. C. M. Rappaport and A. I. Zagloul, Optimized three dimensional
lenses for wide angle scanning, IEEE Trans. Antennas Propag.,
AP-33: 12271236, 1985.
14. R. J. Mailloux, Phased Array Antenna Handbook, Norwood, MA:
Artech, 1994.
15. K. S. Rao et al., Development of a 45-GHz multiple beam antenna
for military satellite communications, IEEE Trans. Antennas Propag., AP-43: 10361047, 1995.
R. B. DYBDAL
The Aerospace Corporation
SCANNING ANTENNAS
SCANNING ANTENNAS
Electronically scanning antennas are usually arrays of elements, with the elements excited in a progressive phase
across the array in a given direction; this scans the beam in
that direction. Such an array is called a phased array. The
array usually produces a narrow (pencil) beam, with modest
or low sidelobes. Arrays that produce a shaped beam are usually not scanned. (Related articles are ANTENNA ARRAYS,
MULTIBEAM ANTENNAS, and ACTIVE ANTENNAS.) Advantages of
phased arrays include the ability to scan and track at rapid
rates, the ability to easily and quickly modify beam shape,
and the flat physical form of the planar array. Mechanically
steered arrays or dishes require a much larger volume. Applications are eclectic, including communication and tracking of
nonstationary satellites, earth mapping synthetic aperture
radar from aircraft or satellite, airport surveillance radar,
ship and aircraft radar and electronic countermeasures, and
finally missile defense.
Important array factors for the systems designer are pattern, gain versus angles, element input impedance, and efficiency. In general, each element of an array will have a different input impedance. For a fixed beam array these are called
embedded impedances; the obsolete and misleading term active impedance is deprecated. Not only does a scanning array
have different element impedances, but each of them varies
with scan angle. These element input impedances are called
scan impedances. The pattern of array gain versus angles is
called scan element pattern; this term replaces active element
pattern. The scan element pattern (SEP) is an extremely useful design factor. The element pattern and mutual coupling
effects are subsumed into the scan element pattern; the over-
685
all radiated pattern is the product of the scan element pattern, and the pattern of an isotropic array of elements that is
scanned to the proper angle. The isotropic array factor incorporates the effects of array size and array lattice, while the
scan element pattern as mentioned incorporates element pattern, backscreen if used, and mutual coupling. Since the scan
element pattern is an envelope of array gain versus scan
angles, it tells the communication system or radar designer
exactly how the array performs with scan, whether blind
angles exist, and whether matching at a particular scan angle
is advantageous. At a blind angle the array has zero gain.
Scan element pattern is used for antenna gain in the conventional range equations. Use of infinite array scan element patterns allows array performance to be separated into this SEP
and edge effects.
Element options for scanning arrays are mostly low gain:
waveguide slots, dipoles, patches, flat TEM horns, and open
waveguides. This is because element spacing of roughly halfwave is needed to obviate grating lobes. Elements such as the
spiral, helix, log-periodic, Yagi-Uda, horn, and backfire are
too large for scanning arrays, except for arrays with very
small scan range. It is important to note that array elements
may be modified for scan impedance compensation, as discussed below.
This article presumes some basic knowledge about antennas; the reader may wish to consult these articles: ANTENNA
THEORY, ANTENNAS, APERTURE ANTENNAS, RADAR ANTENNAS, and
WAVEGUIDE ANTENNAS.
This article comprises four sections: phased array basics,
scanning techniques, mutual coupling effects, and finite
arrays. A more extensive treatment of all of these subjects is
given in the book by Hansen (1).
PHASED ARRAY BASICS
Notation used here is that common in the literature: is
wavelength, k 2/ , and angular variables are u and v
where
u = sin cos sin 0 cos 0
v = sin sin sin 0 sin 0
(1)
F (u, v) =
M/2
N/2
Anm cos n 12 kdx u cos(m 12 kdy v
(2)
n=1 m=1
F(u, v) =
sin 12 Nkdx u
N sin
1
2 kdx u
sin 12 Mkdx yv
M sin 12 kdy v
(3)
Use of low sidelobe designs, such as the Taylor n, or the Taylor one-parameter, requires the Anm coefficient amplitudes to
be properly chosen (2). The coefficients contain the interele-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
686
SCANNING ANTENNAS
(4)
BW
0.866
L sin 0
(9)
for a uniform array. When the beam angle is 30, the commonly used formula for fractional bandwidth results:
Beamwidth
The half-power points on a uniform array principal plane pattern are found by putting Eq. (3) equal to 0.5. This gives,
for all but very small arrays, the 3 dB beamwidth 3 as
arc sin(sin 0 0.4429/Nd)
(5)
(6)
sin 0 sin gl
(11)
1
Grating lobe angle
0.9
0.8
30
45
0.7
60
75
90
(7)
(10)
Grating Lobes
Scan Accuracy
BW =
2
L
d/
BW
0.6
0.5
0
(8)
10
20
30
40
50
60
Scan angle (deg)
70
80
90
SCANNING ANTENNAS
Elements
Phasers
in
Parallel phasers
Series phasers
Figure 2. Linear series feeds.
plane, was developed by Von Aulock (3) and is extremely useful for understanding GL behavior. The GL positions can be
plotted in the uv plane; they occur at the points of an inverse
lattice, that is, the lattice spacing is /dx and /dy. All real
angles, representing visible space, are inside or on the unit
circle. The latter represents 90. Angles outside the unit
circle are imaginary, or in invisible space. When the main
beam is scanned, the origin of the uv plot moves to a new
value, and all GL move correspondingly. However, the unit
circle remains fixed. For further details on this and on hexagonal lattices, see Hansen (1).
GL can be partially suppressed in one plane of scan for
a limited scan range, and over a narrow frequency band, by
pseudorandomizing the position of each element. However,
the complete two-dimensional pattern, over even a modest
bandwidth, will have GL.
SCANNING TECHNIQUES
Phaser Scan
Linear series feeds may be scanned by use of phasers in either the feed line or adjacent to the elements; these are series
phasers and parallel phasers. Figure 2 sketches these arrangements. With waveguide slot array sticks there is no
room for phaser devices. Many other configurations do allow
phasers to be incorporated. An advantage of parallel phasers
is that each phaser handles only 1/N of the transmitted
power, and each element path has the phaser loss. A disadvantage is that the basic interelement phase shift kd sin
0 must be multiplied by the number of the element. Because
almost all phasers are now digitally controlled, this is a minor
disadvantage. Modern digital controls are inexpensive. For
series phasers the advantage is that all phasers have the
same setting equal to . Disadvantages are that the first
phaser must handle almost the entire power and that the
phaser losses are in series. In both topologies there are systematic errors due to impedance mismatch. Amplitude taper
for sidelobe control is provided through coupling adjustments.
A detailed treatment of feed systems is given by Hansen (1).
A word about phasers is appropriate, although a full discussion is outside the scope of this article. Phasers, often
called phase shifters, are primarily of two basic types:
switched line and ferrite. Switched line phasers use semiconductor switches to switch in or out lengths of transmission
line, where the latter is usually coax, microstrip, or stripline.
687
688
SCANNING ANTENNAS
Elements
1
Phasers
Power
dividers
SCANNING ANTENNAS
An alternative to the use of switched series delays is an Nway switch that selects one of N delays. The optical switch
is an acousto-optical Bragg cell, in which an applied microwave signal launches a traveling acoustic wave in the
acousto-optic crystal. An incident optical beam is modulated
by the phononphonon interactions and is deflected by an
angle proportional to the microwave drive frequency. The optical beam then impinges on one of a cluster of optical fibers,
each of which has a binary delay value. Outputs of all the
fibers are collected in an optical power combiner, which in
turn feeds a photodetector (10).
Delay can be selected by choosing the proper light wavelength, through use of dispersive fibers or through use of frequency selective gratings. The RF-modulated optical beam is
sent through a length of highly dispersive fiber; the incremental delay is proportional to the wavelength shift of the laser
carrier (11). A representative chromatic dispersion is 50 ps
per millimeter of wavelength shift per kilometer of fiber. This
dispersion is approximately constant over the laser tuning
range. A variable delay can be obtained simply by tuning the
laser wavelength, with the delay produced by a length of dispersive fiber. A simple (but expensive) array feed would have
a laser oscillator for each array element, with each laser output going through a dispersive fiber and then to a photodetector to the element. Dispersive fiber lengths are of the order of
1 km to 10 km. As mentioned elsewhere, the delay must be
stable and accurate to a small part of the smallest phaser bit
in the array.
A more elegant architecture utilizes one tunable laser (per
plane of scan) with the modulated output fanned out to a cluster of dispersive fibers. These would be designed for delays of
, 2, 4, and (2N 1), and each would drive one array element. Tuning the laser would change and thus scan the
beam. For large arrays the fan-out loss can be appreciable.
For two-dimensional scan, two tunable lasers would be used,
and the azimuth and elevation delays would be combined at
the element photodetectors (12,13).
A grating can be made in a length of fiber by focusing ultraviolet energy onto a periodic sequence of spots along the
fiber, thereby producing a grating of spots of different index
of refraction. Such a grating will reflect an optical wave at
the resonance frequency of the grating. A series of Bragg gratings can be spaced along a fiber, with each reflecting at a
different optical wavelength. An optical signal enters the fiber
with gratings through an optical circulator; the reflected signal path length (delay) depends upon which grating is selected by the optical frequency. The circulator output then
goes to an element, through a photodiode. Thus time delay is
controlled by optical wavelength. It should be noted that the
photonic delay art is relatively recent, while the microwave
feed/phaser art is mature.
Frequency Scan
Advantage can be taken of the beam squint with frequency,
by increasing the interelement path length such that a small
frequency change scans the main beam. Such arrays are traveling wave arrays, as sketched in Fig. 6. Assume that the
radiating elements spaced d apart are connected by a serpertine (snake or sinuous) feed of length s and wave number .
The wavefront is defined by
689
Crossguide
directional coupler
Terminators
Sinuous
feed
Slots
s
d
Linear
arrays
Input
Figure 6. Frequency-scanning sinuous feed feeding an array of traveling-wave linear arrays.
kd sin m = s 2m
(12)
(13)
s
m
dg
d
(14)
690
SCANNING ANTENNAS
N
(N 1)2M (d/) sin 0
(17)
2 sin 0
(18)
Main beam:
First QL:
Second QL:
2
3 4M
(15)
sinc
sin
sin
2
(19)
When /2, the main beam and first QL are equal, clearly
a bad case. But this only occurs for one bit phasers, an extreme case. Table 1 gives lobe amplitudes versus number of
phaser bits.
Gain decrease is approximate given by the main beam decrease:
G
sinc2
Gp
(20)
(16)
Main Lobe
QL1
QL2
1
2
3
4
5
6
3.92
0.912
0.224
0.056
0.014
0
3.92
10.45
17.13
23.58
29.84
35.99
13.46
14.89
19.31
24.67
30.38
36.26
SCANNING ANTENNAS
sin NW
sinc v
N sin W
(21)
m
,
W
m = 1, 2, 3, . . .
(22)
691
(23)
Riso giso (, )
|1 (, )|2
Rg
(24)
4Aelem
cos 0 [1 | (, )|2 ]
2
(25)
where Aelem is the unit cell area of the element. Use of spectral
domain results for slots and dipoles, show that this expression is exact, provided that no grating lobes exist, the
elements are thin and straight, and no higher modes are
engendered by the feed. However, the scan behavior of the
mis-match factor must not be overlooked. It can be shown
that the SEP behavior is closer to cos1.5 , the extra power
being contributed by the impedance mismatch. So it is better
to use the generally applicable formula for SEP to get accurate and useful results. Extensive scan impedance data for
dipole arrays are given by Hansen (1).
Floquets theorem for arrays states that an infinite regular
periodic structure will have the same fields in each cell except
for a progressive exponential multiplier. Furthermore, the
fields may be described as a set of orthogonal modes. In essence the boundary conditions are matched in the Fourier
transform domain, resulting in some cases in an integral
equation reducing to an algebraic equation. Each element is
contained in a unit cell, with all cells alike and contiguous.
The periodic cell approach has proved to be the most powerful
and perceptive technique for understanding and for designing
692
SCANNING ANTENNAS
E plane
D
10
cos1.5
Waveguide Simulators
15
20
0
10
20
30
40
50
degrees
60
70
80
90
5
cos
1.5
10
E plane H
50
80
15
20
0
10
20
30
40
60
70
90
(degrees)
SCANNING ANTENNAS
693
Figure 10. Rectangular lattice waveguide simulators. (Courtesy of Peter Hannan and
Hazeltine Corporation.)
simulators previously shown use circularly polarized or crosslinear polarized elements. E-plane simulation in a diagonal
plane can utilize a pair of modes, TE and TM, with identical
nonzero indices. A detailed review of waveguide simulator design and practice has been given by Wheeler (18). Microstrip
patch arrays have also been simulated. Solbach (19) used two
H-plane waveguide simulators. Here it was important to carefully reproduce the patch feeds and load resistances.
Scan Compensation
Some arrays radiate no power at a certain angle; this is called
a blind angle. This phenomenon can, like array performance
in general, be viewed from the standpoint of either scan reflection coefficient or SEP. The scan reflection coefficient may
develop a magnitude near or equal to unity, or the SEP may
be near or equal to zero. The blind angle occurs when a higher
mode cancels the dominant mode in the element. Since the
relative phase changes rapidly with scan angle, the two
modes produce a resonance, with a resonant phase angle that
can be defined on the decrease of with scan angle. The
higher mode can be produced by external or internal structure. Examples of external structure causing a blind angle are
a dielectric radome covering each element or row of elements,
a dielectric sheet placed on the array face, and a dielectric
plug protruding from the mouth of each element. In all cases
the scan wave impedance at the array face is modified by the
external structure so as to produce a dominant-mode and
higher-mode resonance at a particular scan angle; the higher
mode exists primarily outside the face. Examples of internal
structure that can allow a higher mode that can cause a blind
angle are dielectric plugs flush with the array face, dielectricloaded waveguide elements (as opposed to a short plug), and
a brick array of rectangular guide elements. For these various
structures the waveguide element supports the next higher
mode, which causes the resonance at the blind angle. For the
brick array the higher mode is just below cutoff due to the
large guide size.
A surface wave can be excited on the dielectric substrate
supporting a patch array. The blind angle occurs when the
Floquet scan phase velocity matches the surface wave velocity.
Compensation of scan impedance changes can be made by
any of three different practical methods. These are reduction
of coupling, multimode elements, and external wave transformers.
The simplest way to reduce relative coupling uses H-plane
baffles between slots or dipoles (20). These baffles are simply
metal strips perpendicular to the ground plane and running
between the dipoles or slot rows in the H-plane.
The ideal element power pattern of cos implies that both
TE and TM modes are needed in the array unit cell. The additional modes can be produced by modifying the radiating element in any of several ways. Probably the most effective way
of producing the optimum mix of modes uses open-end waveguide radiators, rectangular or circular, with the waveguide
internal environment suitably modified. A dielectric plug can
be placed at the rectangular waveguide width, and height
steps (in different planes) are also used. The mix of modes
that is needed depends upon the waveguide and lattice dimensions. With a dielectric plug in the waveguide the TE20
694
SCANNING ANTENNAS
of the array) and diminish away from the edge. The Gibbs
amplitudes are given by the Sine Integral, with extrema occurring for arguments of n (24). The period of the oscillations
in scan impedance increases with scan angle, as 1/(1 sin
0), both in the simulated results and in the model. This mode
was derived from the Fourier transform of a single pulse with
a linear phase, and it also utilizes Sine Integrals. Simple finite array models using the Gibbsian form have been constructed, with good comparisons with simulated results (25).
BIBLIOGRAPHY
1. R. C. Hansen, Phased Array Antennas, New York: Wiley-Interscience, 1998.
2. R. C. Hansen, Array pattern control and synthesis, Proc. IEEE,
80: 141151, 1992.
3. W. H. Von Aulock, Properties of phased arrays, Proc. IRE, 48
(10): 17151727, 1960.
4. L. R. Whicker, (ed.), Special issue on microwave control devices
for array antenna systems, IEEE Trans. Microw. Theory Tech.,
MTT-22: 589708, 1974.
5. W. J. Ince and D. H. Temme, Phasers and time delay elements,
In L. Young (ed.), Advances in Microwaves, Vol. 4, New York:
Academic Press, 1969, pp. 1189.
6. S. K. Koul and B. Bhat, Microwave and Millimeter Wave Phase
Shifters, Vols. I and II, New York: Artech House, 1991.
7. E. D. Ostroff, Solid-State Radar Transmitters, New York: Artech
House, 1985.
8. E. D. Cohen, Trends in the development of MMICs and packages
for active electronically scanned arrays (AESAs), In Proc. IEEE
Symp. Phased Array Syst. Technol., Boston, MA, 1996, pp. 14.
9. H. Zmuda and E. N. Toughlian (eds.), Photonic Aspects of Modern
Radar, New York: Artech House, 1994.
10. W. D. Jemison and P. R. Herczfeld, Acoustooptically controlled
true time delays, IEEE Microw. Guided Wave Lett., 3 (3): 72
74, 1993.
11. R. A. Soref, Optical dispersion technique for time-delay beam
steering, Appl. Opt., 31: 78957897, 1992.
12. S. T. Johns et al., Variable time delay of microwave signals using
high dispersion fiber, Elect. Lett., 28: 555556, 1993.
13. M. Y. Frankel, P. J. Matthew, and R. D. Esman, Two-dimensional
fiber-optic control of a true time-steered array transmitter, IEEE
Trans. Microw. Theory Tech., MTT-44: 29692702, 1996.
14. M. S. Smith and Y. C. Guo, A comparison of methods for randomizing phase quantization errors in phased arrays, IEEE
Trans. Antennas Propag., AP-31: 821827, 1983.
15. C. R. Brown and T. F. Carberry, A technique to simulate the
self and mutual impedance of an array, IEEE Trans. Antennas
Propag., AP-11: 377378, 1963.
16. P. W. Hannan, P. J. Meier, and M. A. Balfour, Simulation of
phased array antenna impedance in waveguide, IEEE Trans. Antennas Propag., AP-11: 715716, 1963.
17. R. E. Collin, Foundations for Microwave Engineering, New York:
McGraw-Hill, 1966.
18. H. A. Wheeler, A survey of the simulator technique for designing
a radiating element, In A. Oliner and G. H. Knittel (eds.), Phased
Array Antennas, Proc. 1970 Phased Array Antenna Symp., New
York: Artech House, 1972, pp. 132148.
19. K. Solbach, Phased array simulation using circular patch radiators, IEEE Trans. Antennas Propag., AP-34: 10531058, 1986.
20. S. Edelberg and A. A. Oliner, Mutual coupling effects in infinite
planar arrays. Part I, Slot arrays; and Part II, Compensation effects, IRE Trans., AP-8: 286297, 360367, 1960.
SCHEDULING
21. A. Clavin, D. A. Huebner, and F. J. Kilburg, An improved element for use in array antennas, IEEE Trans. Antennas Propag.,
AP-22: 521526, 1974.
22. B. A. Munk, T. W. Kornbau, and R. D. Fulton, Scan independent
phased arrays, Radio Sci., 14: 979990, 1979.
23. B. L. Diamond, Small arraysTheir analysis and their use for
the design of array elements, In A. A. Oliner and G. H. Knittel
(eds.), Phased Array Antennas, Proc. 1970 Symp., New York: Artech, 127131, 1972.
24. H. S. Carslaw, An Introduction to the Theory of Fouriers Series
and Integrals, New York: Macmillan, 1930.
25. R. C. Hansen and D. Gammon, Standing waves in scan impedance of finite scanned arrays, Microw. Opt. Technol. Lett., 8: 175
179, 1995.
R. C. HANSEN
R. C. Hansen, Inc.
695
SMITH CHART
391
SMITH CHART
The origin of the Smith Chart as a fast and accurate graphical design tool is rooted in the predigital computer era of the
1930s. The original paper chart format was routinely used to
solve difficult transmission line matching problems in a variety of broadcasting, microwave, circuit design, and telecommunications applications. While it has since evolved almost
exclusively into a software format, the original paper forms
remain a convenient detailed medium for print output of finished work from software implementations of Smiths graphical method.
Phillip H. Smith, inventor and developer of the Smith
Chart, was affiliated with Bell Telephone Laboratories for his
entire engineering career after graduating from Tufts College
in 1928. At Bell Labs, Smith repeatedly demonstrated his incisive, pragmatic, problem-solving abilities on a variety of
technical problems. In his first decade on the job he developed
the Chart to its common present form through an iterative
trial process of several more rudimentary forms. Smiths original intended application was to simplify the tedious work of
installing and matching open wire transmission lines for antennas. During this period he almost single-handedly developed and proved the concept of using a graphical chart as a
viable solution method. One can certainly appreciate his desire to minimize tedious slide-rule calculations required to
solve transmission line equations involving hyperbolic functions. Smith (1) cited his routine need to obtain impedance
solutions using Flemings (2) telephone equation as a driving
force to consider a graphical solution alternative. [Readers interested in greater detail on the life of Phillip Smith are referred to the Foreword section of Ref. 3 for a written transcript of the memorial session in Phillip Smiths honor
presented at the 1990 IEEE MTT/S Symposium. An on-line
IEEE source (4) also provides good detail on Smiths background.]
The Chart was first published in a usable developmental
form in 1939 by Smith (5). A subsequent publication (6) in
1944 described the refined form known and used today. A
standard preprinted paper Chart in this format is shown in
Fig. 1(a), and Fig. 1(b) shows only the enlarged central portion around the normalized point z (1 j0). In each of
Smiths publications the fundamental equations and methods
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
392
SMITH CHART
tiv
0
0.11
0.39
Ne
0
11
0.40
0.10
0.4
1
0.0
9
100
100
0.10
0.40
110
0.12
0.38
70
0.
0. 18
32
19
0. 31
0.
40
20
0. 30
0.
30
1
0.2 9
0.2
2
0.2
8
0.2
20
20
0.23
0.27
10
0.24
0.26
0
0.25
0.25
10
0.26
0.24
0.27
0.23
0.2
8
0.2
2
0.2
0. 9
2
1
3
0
32
0. 18
0.
50
0.1
0.3 7
3
3
0.3 7
0.1
60
0
6
0.1
6
0.3
4
4
0.3
6
0.1
70
0.15
0.35
0.35
0.15
0.14
0.36
80
0.36
0.14
80
0.13
0.37
0.37
0.13
90
90
( ZR )
0.38
0.12
9
0.0
1
0.4
Resistance component
0.39
0.11
12
20
8
0.0 2
0.4
X
Z0
ga
07
0. 43
0.
0.4
0.0 2
8
0.
0. 06
44
+J
( (
13
0.
0. 43
07
1
30
44
0. 06
0.
0
0.01 Wave l eng
0.49
0.02 t h s to
0
0.49
0.48
war
0.01
0.48
2
d
0.0 d gen
.0
a
o
7
l
0
d
4
ar
3
i
f
c
f
i
e
e
o
n
w
c
t
0.
era
i
n
n
o
o
i
d
t
t
e
c
g
e
l
r
f
ees
0.4
3 ng t h s g l e o f r e
to
0
.
7
0 ve l e
n
0 . 0r
6
160
a 160 A
4
.
W
0 4
0.4 4
6
0.0
0.
15
0
45
0
0. 05
0. 05
15
P
45
osa
X
J
0.
tiv
er
t Z0
ea
en
14
cta
0
on
4
p
0
nc
1
m
e
co
co
ce
m
n
p
on
ta
ac
en
re
t
0.
0. 31
19
0.
0. 30
20
(a)
Figure 1. Full-size Smith Chart grid. (b) Enlarged center portion of the chart.
1.0
1.1
1.2
1.3
.50
1.0
1.5
2.0
2.5
1.0
1.001
1.002
1.004
1.01
1.02
1.03
1.04
1.05
1.06
90
3.0
Retn Refl.
Loss in db
.20
Vol. Pwr.
.13
.15
.10
.05
.04
Refl. Coef.
.05
1.4
100
0.40
0.10
0.39
0.11
0.38
0.12
0.37
0.13
0.36
0.14
80
0.35
0.15
1.10
0.4
1
0.0
9
0.4
0.0 2
8
0
11
0
70
4
0.3
6
0.1
20
0.
0. 43
07
Capacitive
rea
44
0. 06
0.
,o
iv
ct
du
ri
n
30
(
t
en
pon
ctan
ce c
om
( (
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.
0. 06
44
0.3
1.08
0.27
0.23
1.07
0.26
0.24
.22
.14
15
16
17
18
19
20
.04
.03
3.5
( RZ (, or conductance component ( GY (
0.25
0.25
10
0.3
.20
.18
.16
.02
.01
1.5
60
1.10
( +JB
Y (
.14
.12
.10
4.0
0.1
6
0.3
4
3
0.3 7
0.1
70
0.24
0.26
10
0.2
8
0.2
2
0
0.01 Wave l eng
0.49
0.02 t h s to
0
0.49
0.48
war
0.01
0.48
0.0 d ge
7
0.02w ard lo a d
3 ner
i
f
c
f
i
e
e
o
n
c
t
0.4
in degree
to
ction
e
l
f
s
e
ato
0
r
h
s
f
t
3
.47
.0 e l eng Ang l e o
0
0 . 0r
v
6
1
a
0
4
6
0
W 16
0. 4
0.4 4
0
.
6
0
0.
15
45
50
0
0. 05
0. 05
1
45
0.
JB
Ind
uct
Y0
nt
ive
14
ne
rea
0.7
po
40
0
1
m
cta
o
c
nc
e
c
e
n
c
ta
om
ep
po
sc
ne
su
n
e
0.1
0.23
0.27
20
0.4
.03
.02
.01
Figure 1. (Continued )
0.15
0.35
80
.08
.002
.001
(b)
25
30
40
50
Center
.06
.001
Toward load
.04
Toward generator
90
0.
0. 31
19
Sds Loss
steps coef.
20
.02
In
db
2
0.2
8
0.2
0.2
0.2 9
1
3
0
0.3
tiv
0.14
0.36
0.
0. 30
20
uc
omp
ce c
0.13
0.37
32
0. 18
0.
nd
tan
onent
0 0
Vol.
ratio
1
0.2 9
0.2
30
J
X
Z
X ,
+J 0
Z
i
or
es
ep
usc
0.12
0.38
0.2
20
0. 30
Resistance component
1
20
0.11
0.39
100
0.
40
0.1
07
0. 43
0.
0
13
110
0.4
9
0.0
1
0.4
0.10
0.40
8
0.0 2
0.4
19
0. 31
0.
SMITH CHART
393
0.1
0.3 7
3
0.
0. 18
32
50
394
SMITH CHART
1+k
1k
and k =
z1
z+1
(1)
Table 1. Distributed and Lumped Element Components Commonly Used in Smith Chart Analysis and the Direction of
Progression with Frequency on the Chart
Element Type
Defining Equation
Transmission line
Short-circuit stub
Open-circuit stub
Z in jZ 0 cot l
Inductance
Capacitance
Resistance
E IXL ; XL jL
E IXC ; XC 1/jC
E IR
SMITH CHART
395
l/
Zi
l/
Zi
Z0
Z L=
l/
Z L=
ZL
Z L= 0
Zi
Zi
Z0
Z0
Z L= 0
l/
ZL
l/
l/
(a)
(b)
(Zi + jXL )
Zi
(Zi jXC )
(c)
Figure 2. (a) Concentric circular movement along the chart interior to a value of input impedance, Zi, resulting from addition of a transmission line to a load impedance ZL. (b) Concentric
circular chart movement along the chart periphery for the input impedance of a short-circuit
(ZL 0) and an open-circuit (ZL ) stub. (c) Movement along circular arcs of constant-resistance for addition of a capacitance (downward) and inductance (upward) to a load impedance
ZL.
(hereafter referred to as constant-R and constant-X, respectively). Figure 3(b) uses the same frame to depict loci of constant-k values as solid lined full circles, and it uses loci of
constant phase angle values for variable-k magnitudes as
solid lined partial circles. The constant angle value loci
all intersect at the single point z (1 j0) for which k
(0 j0).
Alternately, one can plot the complex reflection coefficient
[k K exp( j)] in two-dimensional cylindrical (i.e., polar) coordinates as shown in Fig. 4(a), with k (0 j0) located at
the center and cylindrical angle representing the reflection
coefficient angle . Then the transformation to complex normalized impedance using Eq. (1) results in loci for constant-R
and constant-X values which are again full circles and partial
circles, respectively, as illustrated in Fig. 4(b). These circles
and circular arcs have progressively offset centers located
along the orthogonal axes defined by the horizontal line at
x 0 and the vertical line tangent to the outer circle at the
right side point. Figure 4 illustrates the orthogonality features in both domains of (a) reflection coefficient and (b) normalized impedance. The latter grid of circles and circular arcs
represent exactly the Chart solution method first proposed by
Smith to the radio engineering community.
The radii and centers of the constant-R circles and constant-X circular arcs of Fig. 4(b) are derived relative to the
(kxR , kyR ) =
r
,0
r+1
(kxX , kyX ) = 1,
1
x
(2)
(3)
(4)
SMITH CHART
x = X/Z0
+j3
+j2
+j1
VSWR Circles
1
j3
r = R/Z0
j2
VSWR =
j1
x
=
30
+j3
+j2
90
120
60
30
= 0
K =1/3 K =1/2
2
3
4
1 1
j3
j1
60
150
+j1
j2
(5)
(a)
1+K
1K
3
=
00
6
= 360
r
.2
180
.4
.6
.8
K 0
33
150
30
60
120
90
(b)
(a)
x=+j
x=+j2
396
2.
r=
r=
r=
j.5
+ jx
1.
j.5
z=
jx
j.5
j
.1
x=
values of resistance (R) and reactance (X) by the real reference value (Z0). This scaling results in all real axis values to
the left of the center point having values between 0 and 1,
and all real axis values to the right of the center point represent values between 1 and infinity. Charts which plot admittance values are analogously normalized using the reference
value Y0 1/Z0. The usage of normalized impedance values
on the Smith Chart grid facilitates the computation of reflection coefficient, VSWR, and return loss from the chart. Design
examples given in subsequent sections for transmission line
problems illustrate the advantage of using normalized values
for chart-based computation.
(b)
Figure 4. (a) Two-dimensional cylindrical coordinate plot of loci of
constant reflection coefficient magnitude and constant reflection coefficient angle. (b) Loci of constant-R circles and constant-X circular
arcs scaled to fit on the reflection coefficient loci plot of part (a).
SMITH CHART
397
Figure 5. VSWR circle and its associated boundary circle pairs for a
value VSWR 1.25.
VSWR
circle
a
a
a
a
Boundary Circles
The Chart-centered circle for a finite value of VSWR leads
directly to the useful design construct known as boundary circles. They serve as a visual aid to determine the correct value
of a circuit element which will cause the proper amount of
rotation of an impedance or admittance locus for acceptable
matching (i.e., which meets the VSWR requirement).
For a specified VSWR there are two related pairs of boundary circles. Each pair consists of two circles which are centered on the x 0 line and which are doubly tangent to both
the VSWR circle and the periphery of the Chart where the
magnitude of the reflection coefficient is unity. Figure 5 shows
the combination of a VSWR 1.25 circle along with its associated pair of boundary circles. Knowing the directions of rotation for an impedance or admittance locus caused by the addition of series or shunt elements, respectively, one
immediately determines that impedance values over a very
large portion of the Chartnamely, the interior portions of
both small boundary circles, as well as the upper and lower
exclusion areas outside both of the larger boundary circles
398
SMITH CHART
Load Impedance ()
15.0
19.0
21.5
25.0
j250
j227
j192
j180
Chart. To keep the Charts visually succinct, each of the intermediate plotting steps including Z Y and Y Z inversions
as well as the effects of individual branch-by-branch circuit
constructions are presented on a set of individual Charts. The
traces plotted on the Charts use the convention of ZN or YN
for the driving point impedance or admittance when looking
toward the load from branch N. Readers who wish to study a
variety of antenna matching problems executed in step-bystep detail can consult Ref. 13.
In this problem a device, such as an antenna, has a highly
capacitive input impedance vector as given in Table 2 for the
range of 850 MHz to 925 MHz. The objective is to produce a
matching circuit which meets a VSWR requirement of 2.5 : 1
over the reduced frequency range of 855 MHz to 920 MHz
with the restriction that no lumped circuit inductors be used
in the matching circuit.
The bandwidth requirement of about 8.5%, the highly capacitive values, and the implementation restriction of no
lumped inductors all imply that additional elements beyond a
simple two-element L configuration will be required to implement a satisfactory matching circuit. The two solutions derived by use of a software chart method are described in Table
3, giving all the circuit element parameter values. The first
solution uses transmission lines and short-circuit stub elements exclusively and is illustrated in the set of successive
charts in Fig. 6. The second solution requires fewer elements
but includes a capacitance and a short-circuit stub having a
low value of Z0 which is virtually impossible to implement. It
is presented in the second part of Table 3 for purposes of contrast only.
1
2
3
4
a
Element Type
Primary
Transmission line
Shunt short-circuit
stub
Transmission line
Shunt short-circuit
stub
Parameter Values
(v/c group velocity)
Solution a
Z 0 234 , 2.60 cm, v/c 0.69
Z 0 10.1 , 4.84 cm, v/c 0.69
Z 0 76.22 , 0.70 cm, v/c 0.69
Z 0 6.53 , 6.25 cm, v/c 0.69
Contrasting Solution
Transmission line
Z 0 100 , 4.91 cm, v/c 0.69
Shunt capacitance
9.4 pF
Transmission line
Z 0 42.8 , 5.84 cm, v/c 0.69
Shunt short-circuit
Z 0 3.12 , 5.91 cm, v/c 0.69
stub
SMITH CHART
399
Z1
Y1
ZL
(a)
Y2
(b)
Z3
Z2
Y4
Y3
(c)
(d)
400
SMOOTHING CIRCUITS
LEONARD M. SCHWAB
Germaine Development Corp.
SPECTRAL-DOMAIN ANALYSIS
105
SPECTRAL-DOMAIN ANALYSIS
Many numerical methods exist for analyzing microwave and
millimeter-wave passive structures. Among them, spectraldomain analysis (SDA) is one of the most popular. It was developed by Itoh and Mitra in 1974 (1). In 1978, Itoh also presented an SDA version for quasistatic analysis (2). SDA is
basically a Fourier-transformed version of the integral equation method. However, as compared to the conventional
space-domain integral equation method, SDA has several advantages. Its formulation results in a system of coupled algebraic equations instead of coupled integral equations. Closedform expressions can be easily obtained for the Greens functions. In addition, incorporation of physical conditions of analyzed structures via the so-called basis functions is achieved,
and the obtained solutions are stationary. These features
make SDA numerically simpler and more efficient than the
conventional integral equation method. SDA has been used
extensively in analyzing planar transmission lines (e.g., Ref.
3). In this article, we present a detailed formulation of SDA
for planar transmission lines in both quasistatic and dynamic
domains. Other applications of SDA, to resonators and antenna and scattering problems, can be found in Refs. 4, 5, and
6, respectively.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
106
SPECTRAL-DOMAIN ANALYSIS
i (0, y) = i (a, y) = 0
(2)
1 (x, b) = 3 (x, 0) = 0
(3)
2 (x, h3 ) = 3 (x, h3 )
s (x)
2
1
r2
r1
=
y y=h 2 +h 3
y y=h +h
0
2
3
r3 3 y=h r2 2
=0
3
y
y y=h
(1)
a
r1
G1
h1
S1
r2
r3
S2
(6)
(7)
(8)
f (x, y) sin n x dx
(10)
(11)
1 (n , b) = 3 (n , 0) = 0
y
(5)
(4)
(12)
1 (n , h2 + h3 ) = 2 (n , h2 + h3 ) = V U (n )
(13)
2 (n , h3 ) = 3 (n , h3 )
2
1
s (n )
r2
r1
=
y y=h 2 +h 3
y
0
(14)
(15)
y=h 2 +h 3
G2
r3 3
y
h2 b
2
y=h r2
3
y
=0
V U (n ) = V 0 (n ) + V (n )
h3
x
Figure 1. A cross section of the coplanar waveguide used for illustrating SDA.
(16)
y=h 3
s (n ) = s1 (n ) + s2 (n ) + s3 (n )
(17)
(18)
SPECTRAL-DOMAIN ANALYSIS
1 (n , y) = A sinh n (b y)
(19)
2 (n , y) = B sinh n (y h3 ) + C cosh n (y h3 )
(20)
3 (n , y) = D sinh n y
(21)
where A, B, C, and D are unknown functions of n. Substituting Eqs. (19) to (21) into Eqs. (13)to (16) and solving for
U(n) in terms of s(n) yields
V
n ) s (n ) = V U (n )
G(
(22)
N1
n ) = tanh n h
G(
1
r3
tanh n h2 + tanh n h3
r1
0 n tanh n h2 r3
+ tanh n h1 coth n h2
r2
!1
+ tanh n h3 (r1 coth n h2 + r2 tanh n h1 )
Q
C= 0 =
V0
r2
G 1 +S 1 +W
G 1 +S 1
(29)
which can then be used to calculate the characteristic impedance and effective dielectric constant as
Zc =
p1
(30)
c CCa
and
r eff =
d ji s ji (x)
s1i (x) dx
V0
s (x)
d1i
i=1
(23)
j
3
107
(24)
C
Ca
(31)
j=1 i=1
s (n )
j
3
d ji s ji (n )
(25)
j=1 i=1
N1
s1 j G s1i
n=1
i=1
N2
i=1
d3i =
s3 j G s1i
N
3
i=1
d1i +
n=1
2
V
a 0
N2
s2 j G s3i
n=1
n=1
N2
s3 j G s3i
d2i
G 1 +S 1 +W
G 1 +S 1
s1 j (x) dx,
j = 1, 2, . . ., N1
s2 j G s2i
d3i = 0,
i=1
(26)
s1i (x) =
d2i
n=1
d1i +
s1 j G s2i
n=1
i=1
N3
i=1
s1 j G s3i
s2 j G s1i
n=1
i=1
d1i +
n=1
i=1
N2
i=1
N3
N2
j = 1, 2, . . ., N2
s3 j G s2i
(27)
d2i
s3i (x) =
j = 1, 2, . . ., N3
(28)
x S 1 G1
W
2(x S
G1 ) W
1
W
;
x G1
cos i 12
G1
cos
n=1
d3i = 0,
s2i (x) =
cos (i 1)
;
1 (x/G1 )2
1
2
x a + G2
G2
a x 2
G2
2
(32)
(33)
(34)
108
SPECTRAL-DOMAIN ANALYSIS
(35)
G2 " ;
J0
s3i (n ) = (1)i+n+1
i 12 + n G2
2a
;
#
J0
i 12 n G2
p = e, h,
i = 1, 2, 3
(38)
where
(36)
2
(n , y) i2 pi (n , y) = 0,
y2 pi
(37)
i2 = 2 + n2 k2i
(39)
(40)
(41)
(42)
(43)
For 0 x a, y h2 h3,
Ex2 (x, h2 + h3 ) = Ex1 (x, h2 + h3 ) = Ex (x)
(44)
(45)
(46)
(47)
r eff
64
2.7
62
2.6
60
2.5
58
2.4
56
2.3
54
0
10
15
20
25
S2 (mils)
30
35
Characteristic impedance Z0 ()
66
2.2
40
Figure 2. Calculated characteristic impedance and effective dielectric constant of the coplanar waveguide using the quasistatic SDA.
a 100 mil, W S1 2G1 20 mil, h1 h3 4h2 20 mil, r1 1,
r2 2.2, r3 10.5. (1 mil 0.001 in. 0.0025 cm.)
For 0 x a, y 0 and b,
Ex3 (x, y) = Ex1 (x, y) = Ez3 (x, y) = Ez1 (x, y) = 0
(48)
(49)
E z3 (n , h3 ) = E z2 (n , h3 )
(50)
(n , h ) = H
(n , h )
H
x3
3
x2
3
(51)
(n , h ) = H
(n , h )
H
z3
3
z2
3
(52)
E x2 (n , h2 + h3 ) = E x1 (n , h2 + h3 ) = E x (n )
(53)
E z2 (n , h2 + h3 ) = E z1 (n , h2 + h3 ) = E z (n )
(54)
(n , h + h ) H
(n , h + h ) = Jz (n )
H
x2
2
3
x1
2
3
(55)
(n , h + h ) H
(n , h + h ) = Jx (n )
H
z2
2
3
z1
2
3
(56)
E x3 (n , 0) = E x1 (n , 0) = E z3 (n , 0) = E z1 (n , 0) = 0
(57)
E x3 (n , b) = E x1 (n , b) = E z3 (n , b) = E z1 (n , b) = 0
(58)
SPECTRAL-DOMAIN ANALYSIS
3 3
Dh sinh 3 y
E x3 (n , y) = jn De sinh 3 y +
(78)
k
sinh 1 (b y)
E z1 (n , y) = jAe 1
(79)
k2 2
ei (x, y)e j z
Ezi (x, y, z) = j i
(59)
k2 2
hi (x, y)e j z
Hzi (x, y, z) = j i
(60)
i
z t hi (x, y) e j z
(61)
i
z t ei (x, y) e j z
(62)
109
k2 2
E z2 (n , y) = j 2
[Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )]
(80)
k2 2
E z3 (n , y) = jDe 3
sinh 3 y
(81)
(n , y) = jn A cosh (b y) + 1 1 Ae cosh (b y)
H
x1
1
1
h
(82)
(63)
i hi (n , y)
E xi (n , y) = jn ei (n , y) +
(n , y) = jn [B sinh (y h ) + C cosh (y h )]
H
x2
2
3
2
3
h
h
2 2
[Be cosh 2 (y h3 ) + Ce sinh 2 (y h3 )]
(83)
3
3
(n , y) = jn D cosh y
(84)
De cosh 3 y
H
x3
3
h
(64)
(n , y) = jA k1 cosh (b y)
H
z1
1
h
n i
ei (n , y)
hi (n , y) +
E yi (n , y) = j
(65)
k2 2
ei (n , y)
E zi (n , y) = j i
(66)
(n , y) = jn (n , y) + i ei (n , y)
H
xi
hi
(67)
(n , y) = j n i (n , y) + hi (n , y)
H
yi
ei
(68)
t =
x + y
x
y
(n , y) = j
H
zi
k2i
hi (n , y)
(n , y) = jD k3 cosh y
H
z3
3
h
(70)
h1 (n , y) = Ah cosh 1 (b y)
(71)
e2 (n , y) = Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )
(72)
h2 (n , y) = Bh sinh 2 (y h3 ) + Ch cosh 2 (y h3 )
(73)
e3 (n , y) = De sinh 3 y
(74)
1 1
Ah sinh 1 (b y)
E x1 (n , y) = jn Ae sinh 1 (b y)
(76)
E x2 (n , y) = jn [Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )]
2 2
+
[Bh cosh 2 (y h3 ) + Ch sinh 2 (y h3 )]
(77)
(87)
G 11 (n , )E x (n ) + G 12 (n , )E z (n ) = Jx (n )
(88)
G 21 (n , )E x (n ) + G 22 (n , )E z (n ) = Jz (n )
(89)
E x (n )
M
cm E xm (n )
(90)
dk E zk (n )
(91)
m=0
E z (n )
K
k=1
(75)
where Ae,h, Be,h, Ce,h, and De,h are unknown constants. The
fields in the three regions in the spectral domain can now be
derived by substituting Eqs. (70) to (75) into Eqs. (64) to (69)
as
(85)
(69)
e1 (n , y) = Ae sinh 1 (b y)
2
2
(n , y) = j k2 [B sinh (y h ) + C cosh (y h )]
H
x2
2
3
2
3
h
h
(86)
The solutions of Eq. (38) that satisfy Eqs. (57) and (58) are
h3 (n , y) = Dh cosh 3 y
im
P11
( )cm +
m=0
M
m=0
K
ik
P12
( )dk = 0,
i = 0, 1, 2, . . ., M (92)
jk
P22
( )dk = 0,
j = 0, 1, 2, . . ., K (93)
k=1
jm
P21
( )cm +
K
k=1
110
SPECTRAL-DOMAIN ANALYSIS
where
im
P11
=
E xi (n )G 11 (n , )E xm (n )
(94)
E xi (n )G 12 (n , )E zk (n )
(95)
n=1
ik
P12
=
n=1
jm
P21
=
E z j (n )G 21 (n , )E xm (n )
(96)
E z j (n )G 22 (n , )E zk (n )
(97)
n=1
jk
P22
=
n=1
x G S/2
cos
m
[2(x
S/2)/S]
x G W 3S/2
cos m
,
1 [2(x G W 3S/2)/S]2
m = 0, 2, . . .
(98)
Exm (x) =
x G S/2
sin m
1 [2(x G S/2)/S]2
x G W 3S/2
sin m
+
,
1 [2(x G W 3S/2)/S]2
m = 1, 3, . . .
x G S/2
cos k
1 [2(x G S/2)/S]2
x G W 3S/2
k
cos
[2(x
W
3S/2)/S]2
k = 1, 3, . . .
(99)
Ezk (x) =
x G S/2
sin k
S/2)/S]2
1 [2(x G
x G W 3S/2
k
sin
+
,
[2(x
W
3S/2)/S]2
k = 2, 4, . . .
which are defined only over the slots. Their Fourier transforms are found to be
3.5
4th mode
3.0
2.5
2.0
1.5
1.0
0.5
0.0
10
3rd mode
2nd mode
1st mode
15
20
25
Frequency (GHz)
30
35
Figure 3. Propagation constants of the first four modes for the coplanar waveguide calculated using the dynamic SDA. a 60 mil, h1
30 mil, h2 10 mil, h3 20 mil, W 19.7 mil, S1 S2 S 23.6
mil, r1 1, r2 9.6, r3 13.
n S + m
S
cos[n (G + S/2)] J0
4
2
n S m
+ J0
n S + m
S
cos[
(G
+
W
+
3S/2)]
J
n
0
4
2
n S m
+ J0
, m = 0, 2, . . .
2
E xm (n ) =
n S + m
S
(100)
sin[
(G
+
S/2)]
J
n
0
4
2
S m
n
0
n S + m
S
+
sin[
(G
+
W
+
3S/2)]
J
n
0
4
2
S m
J0 n
, m = 1, 3, . . .
2
n S + m
S
j
sin[
(G
+
S/2)]
J
n
0
4
n S m
+ J0
n S + m
S
sin[
(G
+
W
+
3S/2)]
J
+
n
0
4
2
n S m
+ J0
, k = 1, 3, . . .
2
E zk (n ) =
n S + m
(101)
cos[
(G
+
S/2)]
J
n
0
4
S m
J0
n S + m
S
cos[n (G + W + 3S/2)] J0
4
2
S m
, k = 2, 4, . . .
J0 n
2
An example of computed results from the dynamic SDA is
shown in Fig. 3, in which the propagation constants of the
SPEECH ANALYSIS
first four modes for a CPW are plotted against frequency. The
first mode, which is the dominant mode, is found to be propagating up to about 26.5 GHz.
BIBLIOGRAPHY
1. T. Itoh and R. Mittra, Spectral-domain approach for calculating
the dispersion characteristic of microstrip line, IEEE Trans. Microw. Theory Tech., MTT-21 (7): 498499, 1973.
2. T. Itoh and A. S. Hebert, A generalized spectral domain analysis
for coupled suspended microstriplines with tuning septums,
IEEE Trans. Microw. Theory Tech., MTT-26 (10): 820826, 1978.
3. D. Mirshekar-Syahkal, Spectral Domain Method for Microwave
Integrated Circuits, Somerset, UK: Research Studies Press, 1990.
4. T. Itoh, Analysis of microstrip resonators, IEEE Trans. Microw.
Theory Tech., MTT-22 (11): 946952, 1974.
5. Q. Zhang and T. Itoh, Spectral-domain analysis of scattering from
E-plane circuit elements, IEEE Trans. Microw. Theory Tech.,
MTT-35 (2): 138150, 1987.
6. C. Scott, The Spectral Domain Method in Electromagnetics, Norwood, MA: Artech House, 1989.
7. T. Kitazawa and T. Itoh, Asymmetrical coplanar waveguide with
finite metallization thickness containing anisotropic media, IEEE
MTT-S Int. Microwave Symp. Dig., 1990, pp. 673676.
8. K. M. Rahman and C. Nguyen, On the computation of complex
modes in lossless shielded asymmetric coplanar waveguides,
IEEE Trans. Microw. Theory Tech., MTT-43 (12): 27132716,
1995.
9. T. Itoh, Spectral-domain immittance approach for dispersion
characteristic of generalized printed transmission lines, IEEE
Trans. Microw. Theory Tech., MTT-28 (7): 733736, 1980.
10. T. Uwano and T. Itoh, Spectral-domain approach, in T. Itoh (ed.),
Numerical Techniques for Microwave and Millimeter-Wave Structures, Wiley, New York, 1989, pp. 334380.
CAM NGUYEN
Texas A&M University
111
260
SPIRAL ANTENNAS
SPIRAL ANTENNAS
Spiral antennas are so named because of their shape. Often
flat, antenna arms of thin conducting sheet material are defined by spiral curves. Two widely used shapes are related to
the spiral curves shown in Figs. 1 and 2. The spacing between
turns of an Archimedean spiral is constant, but in the logarithmic spiral the spacing grows with the distance from the
origin. An arm of an Archimedean spiral antenna has the spiral curve as centerline and the arm has constant width. An
arm of a logarithmic spiral (log-spiral) antenna has edges defined by two logarithmic spiral curves that differ by rotation
around the origin. Hence, the arm width of a log-spiral antenna also increases with distance from the origin. Usually,
two or four arms comprise an antenna.
The antenna arms cannot follow the spiral curves in the
regions close to or far from the origin. The truncation of the
conductors near the origin provides terminals for connection
to transmission lines or cables. Truncation at the other end,
which defines the length of the arms, is dictated by practical
matters, such as cost, weight, and the available space. The
positions of two arms to form a planar Archimedean spiral
antenna are shown in Fig. 3. The second arm is identical to
the first, but rotated by 180. Figure 4 shows a two-arm logspiral antenna. Note that the small and large limits of this
structure are defined by lines, but could also be defined by
concentric circles of small and large radii centered at the
origin.
Planar spirals can be made by cutting the desired shapes
from thin sheets of good-conducting metal, such as copper or
aluminum. However, a better way is to use photolithographic
techniques like those used to make printed circuits. Thin
sheets of metal-clad microwave substrate (dielectric) material
are commercially available in rather large sizes. The shapes
of the spiral antenna arms are drawn on paper using computer software like that used in preparing Figs. 3 and 4. The
artwork is used to make a photographic negative. The image
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
SPIRAL ANTENNAS
261
Figure 2. Segment of a logarithmic spiral curve. The dot in the center is the origin.
Figure 3. A two-arm, balanced Archimedean spiral antenna. The inner ends are the input terminals.
Figure 4. A two-arm, balanced logarithmic spiral antenna. The outlined edges could be the boundaries of thin metal arms, or of a slot
in a thin metal sheet.
spiral. However, conical spirals are usually designed to radiate mostly in one hemisphere. Depending upon the choice of
the parameters of the spiral curve, the radiation beam of a
conical spiral antenna can be made quite broad or rather
narrow.
In most of the radiation beam of spiral antennas the polarization is circular, or nearly so. The sense of the polarization
is determined by the direction of rotation in which the spiral
expands. Pointing the thumb in the direction of propagation
of the wave radiated by the spiral with the fingers parallel to
the spiral (in the direction of the spiral wrap), the hand used
indicates the sense of polarization following the standards of
the Institute of Electrical and Electronics Engineers. (Physicists often use the opposite designation.) Planar spirals radiate one sense in one hemisphere and the opposite sense in the
other hemisphere. The spiral of Fig. 3 radiates right-hand
sense in the space above the antenna; left-hand sense in the
space below. Spirals with more than two arms must be excited with an arm-to-arm phase progression that agrees with
the sense of the polarization as determined by the direction
of wrap of the spiral. Spiral antennas are sometimes used
even in narrowband applications because of their polarization properties.
One of the first applications of a spiral antenna was on
a navigation satellite. Many other satellite applications have
occurred since then. Circular polarization is preferred over
linear polarization for waves that penetrate the ionosphere
where the plane of linear polarization undergoes rotation.
Conical spiral antennas are used as feeds for reflector antennas. The University of Illinois radio telescope used a linear
array of conical spirals as a slow-speed scanning feed for a
parabolic cylinder reflector (1). The direction of the beam of
such an array can be changed by rotating the antenna elements. This follows from the circular polarization of the spiral
antennas, because the phase of the radiated field is directly
related to the angle of rotation. The broadband capability of
spiral antennas is utilized in direction finding, surveillance,
and electronic warfare systems (25).
262
SPIRAL ANTENNAS
lim
(1)
where is the distance from the origin to a point on the spiral, is the angle measured from a reference axis in the counterclockwise direction, and k is a parameter that determines
how tightly the spiral is wrapped. The sign of k determines
the direction of wrap. The spacing between the turns of an
Archimedean spiral is given by 2k. If Eq. (1) is used to define the centerline of one arm of an antenna, then the centerline of the other arm of a balanced, symmetric antenna would
be given by
= k( )
(2)
(4)
1 d
d
(5)
(6)
Hence, the angle between the radial line and the tangent
is another measure of the spiral rate. Spirals with near 90
are said to be tightly wrapped, but those with small values of
are loosely wrapped. For a given logarithmic spiral, the
angle is fixed, leading to the alternative name, equiangular spiral.
Suppose that the scale is changed on a log-spiral curve, say
by multiplying the radial coordinate by a constant, K 1.
2 = Kea
(7)
2 = ea(+ a
(8)
(3)
1 d
=
= a = cot
d
1
ln K
a
(9)
1 = e0.25
2 = e0.25( /2)
1 = 3
2 = 0.5
and the other one, by
3 = e0.25( )
A
B
4 = e0.25(3 /2)
3 = 2
4 =
3
2
SPIRAL ANTENNAS
= e a
(a)
= k
5
4
3
2
1
(b)
Figure 6. Projection of log-spiral and Archimedean spiral curves in
a plane onto the surfaces of cones. Note that the angle, , between a
radius and the tangent to the spiral curves is constant for the logspiral, but is not constant for the Archimedean spiral. After J. D.
Dyson and P. E. Mayes, New circularly polarized frequency-independent antennas with conical beams or omnidirectional patterns. IRE
Trans. Antennas Propag., AP-9: 4C, 1961.
1 = e(a sin 0 )
2 = e(a sin 0 )( ) = K1
As with the planar log-spiral antennas, rotation of both edges
of one arm will define the edges of a second arm of a balanced
antenna. The angle between the radius vector and the tangent to any edge of the conical log-spiral antenna remains
constant for any cone angle 0.
HISTORY OF DEVELOPMENT
Experiments with Archimedean spiral antennas were begun
in 1953 by E. M. Turner at Wright Air Development Center,
Dayton, Ohio (6). The problem of limited antenna bandwidth
was hampering the development of broadband homing, direction-finding and other electronic systems of interest to the
U.S. Air Force. The Research Laboratory of Electronics at
Massachusetts Institute of Technology (MIT) was awarded a
263
contract to develop a theory for the Archimedean spiral antenna and much development work was done at the Naval
Research Laboratory, Diamond Ordnance Fuze Laboratories,
and elsewhere. Other contracts went to the Antenna Laboratory of the University of Illinois at Urbana-Champaign
(UIUC) to study the general problem of greatly increasing antenna bandwidth.
In September of 1954, V. H. Rumsey came to the UIUC
Antenna Laboratory from Ohio State to assume the position
of director. It was well known that the performance of an antenna was determined by its size in wavelengths. Rumsey had
already tested some preliminary ideas about antennas that
were described, insofar as possible, by angles rather than
lengths. He also recalled that Mushiake had pointed out that
a self-complementary antenna would have a constant input
impedance, independent of frequency (7). Little practical use
had been made of this principle because all self-complementary shapes extend to infinity. It occurred to Rumsey that an
unbounded logarithmic (equiangular) spiral could be specified
completely by a single angle and it could be used to define a
self-complementary antenna (8). A two-arm, log-spiral antenna is said to be self-complementary when the arm edges
are spirals that are 90 apart (as are those in Fig. 4). The
space between arms is then congruent to the arms, the condition to be self-complementary.
But could a physical antenna, necessarily a truncated version of the ideal, be made to operate like the infinite one, even
over a finite band? He suggested to John Dyson, a graduate
student in the Antenna Laboratory, that self-complementary
antennas using log-spirals as boundaries should be investigated as potential broadband antennas.
It was not readily apparent how to isolate the performance
of the spiral-shaped antenna from the effects of the cable that
must be attached to connect the antenna to transmitters or
receivers for testing. By making the spiral arms as a slots in
a large ground plane, rather than a thin conducting sheet, the
cable could be affixed to the ground plane where it would
cause minimum perturbation of the field. So the terminals of
the spiral slots were extended directly toward one another to
meet at the origin, the shield of the feed cable was soldered
to the area between the spiral slots up to one side of the slot
near the origin, and the center conductor was attached on the
other side of the slot. In some instances, a dummy cable was
soldered in a symmetrical location to improve the balance of
the feed region and the symmetry of the radiation patterns.
This feed provides a simple conversion from the unbalanced
cable to the balanced antenna, the function of devices called
baluns.
Dyson demonstrated that planar log-spiral slot antennas
in the shape shown in Fig. 4 (with the terminals connected
by continuation of the slot) could indeed be made to have impedance and pattern characteristics that changed very little
over bandwidths greater than 10:1. Dyson later studied planar sheet log-spirals and conical log-spirals and gave design
and application information (9).
THEORY OF OPERATION
As with other antennas, the distribution of current produced
by applying a source of radio frequency power to the terminals is the key to understanding the performance of spiral
264
SPIRAL ANTENNAS
SPIRAL ANTENNAS
=0
f = 595 MHz
r = 1.94
= 90
265
E
E
f = 800
r = 1.39
f = 1000
r = 1.19
f = 1400
r = 1.04
E
E
Figure 7. Measured radiation patterns of a two-arm, balanced, logspiral antenna, r axial ratio. (After J. D. Dyson, The equiangular
spiral antenna. IRE Trans. Antennas Propag., AP-7: 2C, 1959.
=0
f = 2800 MHz
r = 1.05
= 90
;;;;
Z
f = 3950
r = 1.07
d
2
f = 3600
r = 1.07
f = 9027
r = 1.07
f = 12000
r = 1.24
H
h
d
2
E
E
Figure 8. Measured radiation patterns of a two-arm, balanced, logspiral antenna, r axial ratio. After J. D. Dyson, The equiangular
spiral antenna. IRE Trans. Antennas Propag., AP-7; 2C, 1959.
Figure 9. Parameters of a conical log-spiral antenna and the coordinates used to describe radiation patterns. After J. D. Dyson, The
characteristics and design of the conical log-spiral antenna, IEEE
Trans. Antennas Propag., AP-13: 7C, 1965.
SPIRAL ANTENNAS
E
E
= 90
o = 10
o = 15
o = 30
o = 90
Figure 10. Radiation patterns of a conical log-spiral antenna showing increasing front-to-back ratio with decreasing cone angle, 0. After
J. D. Dyson, The unidirectional equiangular spiral antenna. IRE
Trans. Antennas Propag., AP-7: 4C, 1959.
fire. As frequency changes, the active region moves, maintaining constant size in wavelength measure.
When the antenna is wrapped more loosely ( 60), the
active region is broader and includes turns that are phased
away from backfire toward broadside. This accounts for the
broader beams. This effect becomes even more pronounced as
the wrap is loosened further ( 45). Fluctuation in the
near-field amplitude in Fig. 11 is caused by the probe passing
close to conductors of the spiral. The dashed lines show
smoothed data more indicative of the behavior of the near
field along the arm rather than that along a radial line.
The plots of phase versus displacement in Fig. 11 illustrate
how the phasing is affected by the increasing length of conductor in each successive turn of the log-spiral. Appreciable
slope is seen in the active region for 80. As the slope in
the active region decreases, more turns are included in the
active region and some of these turns are phased in directions
slightly removed from backfire. The bandwidths of antennas
with wide active regions are smaller for a given physical size
than those of antennas with more limited active regions.
Apparently because of the effective small wrap angle of
conical log-spiral antennas, the pattern performance does not
depend greatly upon the width of the arms. In fact, for moderate bandwidths it is possible to eliminate the flat conductors
altogether if a dummy cable is used on the second arm as
recommended to preserve the symmetry.
=0
Active region
Phase
1000
0
3
900
= 80
800
700
10
15
Relative amplitude in dB
266
0
500
0
.5
1.0
1.5
2.0
2.5
2.0
2.5
400
= 60
300
3
10
15
0
200
.5
1.0
1.5
100
0
= 45
3
10
15
30
1.5
.5
1.0
2.0
Distance along conical surface ( / )
2.5
Figure 11. Relative amplitude and phase of magnetic fields (currents) measured along the surface of conical log-spiral antennas
(20 15, 90) and corresponding radiation patterns. After J. D.
Dyson, The characteristics and design of the conical log-spiral antenna. IEEE Trans. Antennas Propag., AP-13: 7C, 1965.
SPIRAL ANTENNAS
Several parameters are required to describe a single conical log-spiral antenna. The pattern shape is most dependent
upon the apex angle, 20, of the cone. However, the arm
width, , and the wrap angle, , can also have an effect. Radiation patterns for several values of 20, , and are shown in
Fig. 12. As to be expected, tightly wrapped antennas have the
smoothest, most symmetrical patterns for a wider range of
20 and . However, the front-to-back ratio is best for the
smallest apex angle, 20 15. Antennas with arm widths
given by 90 have patterns superior to those with either
wider or narrower arms. Infinite planar spiral structures with
90 are self-complementary. Although the argument for
constancy of impedance applies only to such planar cases, evidently there are other advantages to be gained by applying
the self-complementary condition.
The measured input impedance does not change much over
the frequency band of good patterns, a span of 20 : 1 or more
in frequency is possible. Table 1 gives data for the mean impedance and SWR as a function of the cone angle, 0.
Multiarm Log-Spiral Antennas
Although most log-spiral antennas in use today have two
arms, that is by no means a requirement. The needs of certain
applications can be better met with an antenna having more
than two arms. Log-spiral antennas with only one arm can be
fed against ground, but the performance is unsuitable for
most applications (14,15). Two-arm log-spirals are conventionally fed with 180 degree phase difference between the
arms at the input. As the number of arms increases, other
= 60
= 70
= 80
= 0 = 90 = 0 = 90 = 0 = 90
267
Max. SWRa
10
15
30
90
129
147
153
164
1.9
1.9
1.95
2.1
a
Referred to the mean impedance.
Source: J. D. Dyson, The unidirectional spiral antenna, IRE Trans. Antennas
Propag, AP-7: 4C, 1959.
possibilities arise, as shown in Fig. 13. Some of these are simply related to the azimuthal ( dependent) solutions of Maxwells equations, exp(jm), where m is an integer to ensure
single-valued fields. The conventional feed for two-arm spirals
can produce fields having only odd integer values of m, because a rotation of 180 is in agreement with a phase change
of 180. However, measurements of the fields radiated from
conventionally fed two-arm spirals find that contributions
from values of m greater than one are negligibly small.
Hence, it might be expected that a field which varies as
exp(j2) could be produced by feeding a four-arm spiral as
shown in Fig. 13(c). Recalling that the sense of polarization is
determined by the direction of wrap, trying to excite a
multiarm spiral in the wrong sense leads to poor performance. The excitation of Fig. 13(c) applies equally well to
m 2, but the direction of wrap determines the sense of
the polarization. The purity of the phase law as given by
exp(jm) is demonstrated by the measured phase of the
= 16
= 90 2 o = 15
+
(a)
= 164
= 16
+1
1
= 90 2 o = 20
+1
1
= 164
(b)
= 16
1
= 90 2 o = 30
+1
+1
1
= 164
(c)
Figure 12. Typical radiation patterns indicating changes in shape
with cone angle, 0, spiral angle, , and angular arm width, . After
J. D. Dyson, The characteristics and design of the conical log-spiral
antenna. IEEE Trans. Antennas Propag., AP-13: 7C, 1965.
SPIRAL ANTENNAS
E
E
= 73
= 60
= 45
40
50
in degrees
268
60
70
80
90
40
60
70
50
in degrees
80
90
720
680
+1
640
1
600
Planar spiral antennas in free space find little application because of the equal radiation on both sides of the plane. If the
application will allow 3 to 4 dB loss, the power radiated from
+1
560
520
480
440
Archimedes spiral
Equiangular spiral
400
sin 10
(
)
= e tan 45
= 1.026
360
320
f = 1000 MHz
E
E
280
240
200
f = 1400 MHz
160
120
+1
80
f = 2000 MHz
40
0
40
(a)
(b)
Figure 16. Radiation patterns of symmetrical, four-arm (a) Archimedean and (b) equiangular spiral antennas; 0 10, D 29.5 cm,
d 4.5 cm, ( 90, variable). After J. D. Dyson and P. E. Mayes,
New circularly polarized frequency-independent antennas with conical beams or omnidirectional patterns. IRE Trans. Antennas Propag.,
AP-9: 4C, 1961.
SPIRAL ANTENNAS
one side can be absorbed. This is often accomplished by placing a conducting cavity on one side of a planar spiral and
introducing resistance cards, lossy foam, or some other absorbing material into the cavity. Sometimes the losses can be
avoided if the band to be covered is sufficiently narrow and
some degradation in performance can be permitted.
Some recent publications have indicated that spiral antennas can operate well when placed parallel to a closely spaced
conducting ground plane (16). The effect of the close ground
is to reduce the rate of decay of the currents on metal spiral
arms. Hence, it may be necessary to absorb the power that is
left at the truncation to prevent large reflected power from
degrading the pattern and impedance. A ring of absorber
around the outer rim of a spiral antenna may dissipate less
power than an absorbing cavity and provide thereby increased gain.
An alternative to the peripheral absorber ring is a vertical
resistance card that spans the space between the spiral arms
and the ground plane and follows along an edge or the centerline of the spiral. This method has the potential to control the
rate of decay of the currents so that negligible current remains to be reflected at the truncation. Although computer
simulations have shown the card-loaded spirals over close
ground to work well, the construction of such an antenna is
difficult and the measured performance is not as good as predicted (15).
The electromagnetic boundary-value problem presented by
a log-spiral antenna has, so far, proved intractable. Analytic
results have been obtained for anisotropic conducting sheets
with spiral lines of conductivity (17,18), but this model is
somewhat divergent from the practical case. The availability
of high-speed digital computers has made it possible to solve
discretized versions of integral equations for practical log-spiral geometries. The conducting arms of a spiral antenna are
divided into many small areas (patches). The current on each
patch is represented by some (basis) function with an unknown complex-valued coefficient. Expressing the field in
terms of these coefficients enables one to convert the integral
equation into a set of linear algebraic equations that can be
solved with standard techniques. The digital computer is capable of solving such systems of equations that involve thousands of unknowns. The coefficients so determined can be
used with the basis functions to calculate an approximate current distribution on the antenna. The calculation of the radiated fields, polarization and patterns, from the approximate
currents is straightforward. An approximate value for the input impedance can be determined from the value of the current at the input terminals. Additional information about
applying these method of moments techniques to spiral antennas can be found elsewhere (19,20).
Finding the currents on multiarm spirals, particularly
those used with closely spaced ground planes or cavities (with
or without absorber) and intended for use over very wide
bands, can tax the capability of even the fastest supercomputer. One way to reduce the number of unknowns on many
spiral antennas is to make use of the discrete rotational symmetry, the fact that each arm is an exact duplicate of the
others and the separations between adjacent arms are the
same. Further work in numerical analysis will likely yield
faster algorithms for computing the electromagnetic fields
produced by various distributions of currents and this will
reduce the computer time needed to solve problems with large
numbers of unknowns. Alternative methods for evaluating
269
270
SPONTANEOUS EMISSION
PAUL E. MAYES
University of Illinois at UrbanaChampaign
562
TELEVISION ANTENNAS
TELEVISION ANTENNAS
At the present time, the superturnstile antenna (of which the
batwing antenna is a radiating element) (1), the supergain
antenna (dipole antennas with a reflector plate), invented in
the United States, and the Vierergruppe antenna, invented
in Germany and sometimes called the two-dipole antenna in
Japan, are widely used for very high frequency television
(VHF-TV) broadcasting around the world. The superturnstile
antenna (2) is used in Japan, as well as in the United States.
Figure 1 shows the appearance of the original batwing antenna, when it was first made public by Masters (1). The secret lies in its complex shape. It is called a batwing antenna
in the United States and Schmetterlings Antenne (butterfly
antenna) in Germany, in view of this shape.
The characteristics of the batwing antenna were calculated
using the moment method proposed by Harrington (3) and
experiments were conducted on this antenna. The model antenna was approximately two-fifths the size of a full-scale batwing antenna, with its design center frequency at 500 MHz.
It has been reported that for thick cylindrical antennas, a
substantial effect on the current distribution appears due to
nonzero current on the flat end faces. The assumption of zero
current at the flat end face is appropriate for a thin cylindrical antenna; however, in the case of a thick cylindrical antenna, this assumption is not valid.
Specifically, this article applies the moment method to a
full-wave dipole antenna, with a reflector plate supported by
a metal bar, such as those widely used for TV and frequency
modulated (FM) broadcasting (4). In the present study, the
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
TELEVISION ANTENNAS
P( . )
E ( )
563
E ( )
3.2
12.2
3.1
A
I8
H
I7
B
I6
G
37.8
I5
D
I4
I3
E
I11
I10
I2
I1
I9
300 MHz
(a)
Jumper
72 Branch cable
l
(Unit: cm)
Figure 2. Construction of the model batwing antenna and its coordinate system.
500 MHz
(b)
700 MHz
(c)
Figure 3. Amplitude characteristics of current distribution for frequency range from 300, 500, 700 MHz of shaded areas.
564
TELEVISION ANTENNAS
1.0 0
1.0
1.0
E ( )
Amp.
0.5
E ( )
0
200
90
200
0
phase (deg.)
180
300 MHz
0.5
0.5
Horizontal pattern
700 MHz
500 MHz
(a)
(b)
(c)
1.0 0
1.0
1.0
Theoretical values
Measured values
(amplitude pattern)
Theoretical values
Measured values
(phase pattern)
E ( )
0.5
E ( )
0
200
90
0
200
phase (deg.)
180
300 MHz
0.5
0.5
Vertical pattern
700 MHz
500 MHz
(a)
(b)
(c)
Figure 4. Amplitude and phase characteristics of radiation patterns at 300, 500, 700 MHz (with
support mast).
Normalized impedance 72
Theoretical
value
Input resistance
100
4
Measured
value
50
Input reactance
0
300
400
500
3
Gain (dB)
Input impedance ()
150
600
700
Frequency (MHz)
50
100
Figure 5. Theoretical and measured values of the input impedance
as function of frequency (mast is infinite thin, 0).
0
300
400
500
Frequency (MHz)
600
700
TELEVISION ANTENNAS
565
2h
2as
h/2
Metal-bar
supported
lr
2a
Dipole
antenna
#5
#6
#4
#2
#0
Reflector #1
plate
=
=
=
=
=
=
=
=
#3
2as
(a)
a/ o
a/ o
s/ o
d/ o
d/ o
lr/ o
lt/ o
as/ o
0.04
0.01
dr
0.4
0.8
0.15
1.2
0.2 or 0.25 or 0.3
0.01
lt
2a
2ar
(b)
TELEVISION ANTENNAS
f/f0
0.6
0.6
#3
0.4
0.6
#4
0.4
0.2
#5
0.4
#6
0.35
0.3
0.2
0.2
0
180
0
180
0.6
#2
0.5
180
0
180
(mA/V)
0.5
#0
2
#1
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
180
0.6
180 3
0
180
0.6
#3
0.4
0
180
0.6
#4
0.4
0.2
0.1
0.7
0
180
0.35
0.3
0.2
0
180
0
180
0.6
#2
#1
0.6
#6
0.4
0.2
#0
180
#5
180
0.1
180
0.5
0.4
0.2
0
180
0.6
#3
0.4
0.4
0.2
0.2
0
180
0.2
2
#4
0.6
0.4
0
180
0.6
#5
0.4
0.4
0.2
0.2
#6
0.35
0.3
0.1
180
1.0
566
0
180
0
180
2
#0
0
180
2
0.6
#1
0.6
180
#2
0.5
0.4
0.4
0.4
0.2
0.2
0.2
0
180
0
180
0
180
Figure 9. Current distribution of metal-bar supported full-wave dipole antennas (two-bay) with
a wire screen-type reflector plate.
TELEVISION ANTENNAS
1.0
0.9
0.8
0.7
0.6
0.5
f/f0
l1/ 0 = 0.2
l1/ 0 = 0.25
l1/ 0 = 0.3
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180
180
(
Amplitude
567
180 0
0 180
180
Phase)
Figure 10. Radiation pattern of metal-bar supported full-wave dipole antennas (two-bay) with
a wire screen-type reflector plate.
568
TELEVISION ANTENNAS
300
Re
200
lt
Zin ()
100
Im
Parallel line
0.6
0
0.7
f/f0
0.5
0.8
0.9
Loop
Trap
0.2
0.25
0.3
An infinite reflector
200
10
Gain (dB)
l1/ 0
0.2
0.25
0.3
4
0
2l1
l1/ 0
100
1.0
0.5
0.6
0.7
f/f0
0.8
0.9
1.0
TELEVISION ANTENNAS
l1 = 0.15 0
0.4
600 MHz
l1 = 0.25 0
0.4
0.2
0.2 0.4 0.6 0.8
=
90
1.0
0.2
0.4
0.4
=
90
1.0
=
90
1.0
=
90
1.0
=
90
1.0
180
0
0.2
0
=
90
1.0
180
0
0.4
=
90
1.0
0.2
0
0.2
0.2
0.4
0.4
0.6
180
0
0.2
0.4
0.6
180
0
0.2
0
=
90
1.0
0.2
0.4
900 MHz
0.4
0.4
180
0
0.2
0.4
0.6
180
0
0.2
0.2
825 MHz
0.2
0.4
0.4
180
0
0.2
0.4
0.4
=
90
1.0
0.2
0.2
750 MHz
0.2
0.4
0.4
180
0
0.2
0.4
0.4
0.2
0.2
675 MHz
569
180
0.6
Theoretical
=
90
1.0
180
Measured
frequency. It is very simple, with one feed point per bay and
a maximum of four power-dividing elements for an eight-bay
array. The radiating elements are grounded for maximum
lightning protection. Figure 22 shows the four bay V antenna
pattern for various values of space.
CONCLUSION
Previous researchers (1,2) have analyzed batwing antennas
by approximating the current distribution as a sinusoidal distribution. Wideband characteristics are not obtained with a
sinusoidal current distribution. In this article, various types
of modified batwing antennas, as the central form of the superturnstile antenna system, were analyzed theoretically
with the aid of the moment method. The results were compared with measurements, in order to examine the performance of the antenna elements in detail.
570
TELEVISION ANTENNAS
600 MHz
= 90
0
0.4
#1
#2
#3
#4
#5
0.15 0
0.8
= 90
1.0
0.4
0.8
1.0
675 MHz
= 90
= 90
S
0.4
0.8
1.0
0.8
0.4
1.0
3 0
l2
750 MHz
l1
= 90
0.4
0.8
= 90
1.0
0.4
0.8
1.0
825 MHz
#17
#18
#19
#20
#21
= 90
0.4
0.8
= 90
1.0
10
0.4
0.8
1.0
900 MHz
= 90
0.4
Slot
Slot
0.8
= 90
1.0
0.4
0.8
1.0
An infinite reflector
(a)
(b)
Probe
0
30
330
Probe position
60
300
Slot
Probe
Relative field
20 40 60 80
270
90
Excitation of slot
(b)
240
120
Construction
(a)
Figure 17. The traveling-wave slot antenna.
210
150
180
TELEVISION ANTENNAS
571
0.235
A triangular
prism
0.340
Diameter
/100
Pitch
p = /2
Feed point
A cylindrical conductor
Short
(a)
(b)
Freq.
Pattern
470 MHz
480 MHz
Horizontal
pattern
180
90
0 180
270
90
Vertical
pattern
2nd mode
90
180
270
90
270
Horizontal
pattern
Vertical
pattern
3rd mode
0 180
270
90
180
0 180
0 180
270
270
90
0 180
270
90
0 180
270
730 MHz
90
0 180
0 180
270
0 180
270
270
90
270
90
720 MHz
90
270
90
0 180
0 180
710 MHz
90
90
270
90
270
700 MHz
90
0 180
0 180
510 MHz
90
270
90
270
690 MHz
500 MHz
90
0 180
0 180
180
490 MHz
270
90
0 180
270
270
572
9 10
0
/
0=180
120
20
10
BIBLIOGRAPHY
30
2. Y. Mushiake, (ed.), Antenna Engineering Handbook. The OHMSha, Ltd., 1980 (in Japanese).
3. R. F. Harrington, Field Computation By Moment Method, New
York: Macmillan, 1968.
4. W. Berndt, Kombinierte Sendeantennen Fur Fernseh-und UKWrunfunk (Teil II), Telefunken-Zeitung, Jahrgang, Heft. 26, 101:
268, 1953.
60
90
5. Y. T. Lo and S. W. Lee, Antenna Handbook, New York: Van Nostrand Reinhold, 1988.
0
0.5
270
90
0.5
HARUO KAWAKAMI
Antenna Giken Corporation
0.5
180
0
0.75
270
90
0.75
0.75
180
0
1.0
270
90
1.0
1.0
180
Figure 22. Vertical and horizontal pattern of V-shaped antenna.
that the change in the characteristics with a change in frequency becomes more severe with increasing trap length lt,
and the bandwidth becomes small, while a short trap length
lt shows a small change and a tendency for the bandwidth to
become wide. For lt 0, a wide bandwidth for pattern and
gain was obtained for the 2L type. The input impedance has
a value very close to 50 , essentially the same as the characteristic impedance of the feed cable over a very wide frequency range. Thus, a satisfactory explanation was given
with regard to the matching conditions. Popular television antennas cover the properties of many basic types of antennas
which are the mainstream of antenna technology. It has been
492
WAVEGUIDE ANTENNAS
WAVEGUIDE ANTENNAS
Waveguide antennas are useful for a variety of airborne,
ground based, space, and missile applications in the microwave frequency band because they offer high radiating efficiency, a high degree of control over the radiation pattern,
compact architecture, and low weight. The high radiating efficiency and compact architecture, namely, the small depth
compared to a parabolic reflector antenna with equivalent
gain, combine to optimize the use of the available aperture
area. Accurate control of the radiation pattern allows designing arrays with a radiation pattern envelope tailored to the
specific application. The low weight minimizes the impact on
the installation and allows mounting the antenna on a mechanical gimbal assembly to scan the radiation pattern. A
waveguide slot array designed to mount on a gimbal for an
airborne radar application is shown in Fig. 1.
In waveguide slot arrays the radiating and transmission
structures are inexorably intertwined, thus it is appropriate
to treat the history of the waveguides together with that of
slot radiators. In the 1930s, G. C. Southworth (1) at Bell Telephone Laboratories investigated electromagnetic propagation
in hollow waveguides. Carson, Mead, and Schelkunoff performed the accompanying theoretical analysis (2). Independent work by Barrow of MIT (3) was published at about the
same time (4). The analysis of waveguides and slot radiators
in terms of transmission lines is also due to Schelkunoff (5).
Another aspect of array work is the need for pattern synthesis. S. A. Schelkunoff s classic 1943 paper (6) and the subsequent C. L. Dolph (7) paper on the synthesis of equal sidelobe
levels were used by T. T. Taylor (8) to achieve practical array
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
WAVEGUIDE ANTENNAS
E s (x, z) = x
Vs
cos(z/l) for w/2 < x < w/2, l/2 < z < l/2
w
(1)
where Vs is the peak voltage across the slot. The far-field radiation pattern of the slot located in an infinite ground plane is
calculated by using Babinets principle (15) (or image theory).
For a slot located in the xz plane, this results in a radiation
pattern given by
e jkR
R
jV s k sin cos(kl/2 cos )
E (, ) =
l
(/l)2 k2 cos2
E = E (, )
(2)
jV s cos(/2 cos )
sin
(3)
D(, ) =
2
0
4P(, )
P(, ) sin d d
(4)
1
2
r
|E(, )|2
493
(5)
Hz = cos
mx
a
cos
ny
b
e mn z
mx ny
mn
mmn
Ey = 2
sin
cos
e mn z
j
mn a
a
b
mx ny
mn
nmn
Ex = 2
cos
Hy =
sin
e mn z
j
mn b
a
b
m 2 n 2
2 =
+
a
b
Hx =
(6)
mx
ny
e mn z
b
mx ny
mn
mmn
Hy = 2
cos
Ex =
sin
e mn z (7)
j
mn a
a
b
mx ny
mn
nmn
Hx = 2
sin
Ey =
cos
e mn z
j
mn b
a
b
Ez = sin
sin
r m
n 2
b
k2
(8)
= 2/
(m/a)2 + (n/b)2
(9)
494
WAVEGUIDE ANTENNAS
Hz = j cos
x
a
e j 10 z
x
10
sin
e j 10 z
/a
a
x
0
sin
Ey =
e j 10 z
/a
a
Hx =
(10)
Ys
where the upper sign denotes a wave traveling in the zdirection and the lower sign denotes a wave traveling in the
z-direction. The propagation constant for this mode is given
by
10 = j10 = j
k2
2
a
(11)
g =
(12)
2a
The relationship J 1n H gives the electric currents flowing on the broadwall surface of the waveguide for this mode
as
Jx = j cos
Jz =
x
a
10
sin
/a
e j 10 z
x
a
e j 10 z
(13)
Figure 3. Waveguide radiating slots and their lumped-element circuit models: offset longitudinal shunt slot, and angled and transverse
series slot.
B10 = A10 =
Zs
sin
x
a
(15)
WAVEGUIDE ANTENNAS
Vs
Y
= K1 f (x, l)
G0
V
2
kG0 10 ab
(16)
and
f (x, l) =
j2V
cos
0 (10 /k)ab
10
2k
sin
x
a
Offset=0.5 cm
0.6
Length=1.6 cm
Length =
1.5 cm
0.5
Offset=0.38 cm
0.4
0.3
Offset=0.25 cm
0.2
0.1
0
8.25
Offset=
0.13 cm
8.50
8.75
9.00
9.25
9.50
9.75
(17)
and the slot is represented by a shunt conductance. The conductance value is a function of the slot offset and the waveguide dimensions. The dependence of the conductance on the
slot offset is given by (10,18)
G
x
= g = g1 sin2
G0
a
g a
g1 = 2.09
cos2
b
2g
Length=1.7 cm
0.7
(18)
These equations are useful for understanding the slot characteristics but are not accurate enough for many design applications. The amount of radiation from the slot is controlled primarily by the offset and length as expressed previously but
also depends on the slot width and the thickness of the wall
in which the slot is cut. The slot resonant frequency must also
be determined accurately. The slot is defined to be resonant
when the phase of the scattered field is 180 out of phase with
the incident wave. The resonant frequency occurs when the
slot length is approximately one-half of a free-space wavelength, but the exact frequency is also a function of the slot
offset, width, and thickness. The characteristics of the slot
are determined more accurately by using an electromagnetic
simulation technique, such as the method of moments (19), or
by careful measurements. Figure 4 shows the variation of the
slot conductance and resonant frequency as the offset and
length are varied. The slot characteristics were calculated by
using a computer simulation based on the method of moments
formulation described in Khac (20), with slot thickness effects
were added. The variation of the slot admittance normalized
to the peak conductance values versus frequency relative to
the resonant frequency for a slot in full height (2.3 cm 1.0
cm) X-band (8 GHz to 12 GHz) waveguide is shown in Fig. 5.
Away from the resonant frequency, the slot becomes reactive,
and the radiation is reduced. The variation with frequency is
relatively insensitive to the slot offset and length, so the
curves are applicable over a wide range of offset and length
0.8
1
0.9
g/g0
b/g0
0.8
0.7
0.6
0.4
0.6
0.2
0.5
0.4
0.3
0.2
2
K1 =
ja
where
0.8
Normalized slot conductance at resonance
495
0.2
0.1
0.4
0
0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04 1.06 1.08 1.10
Normalized frequency f/f0
Figure 5. Variation of normalized conductance and admittance for
an offset-shunt slot relative to the resonant frequency.
496
WAVEGUIDE ANTENNAS
x
R
1
= r = r1 cos2
R0
2
r1 = 0.523
3
2
cos2
ab
4a
(19)
The maximum coupling for this element occurs when the slot
is centered. This slot has a high coupling value, but the range
of coupling attainable is limited because the slot length restricts the range of offsets, and the slot is confined to the
broadwall.
A slot located along the waveguide centerline also radiates
when rotated at an angle relative to the centerline. The forward scattering from this element is opposite in phase to that
of the backscattered field, and hence it is modeled as a series
lumped element. The amount of power coupled is controlled
primarily by the slot angle and is adjustable over a wide
range of coupling values.
2
R
= r = 0.131
R0
cos
I( )
2
=
1 2
J( )
=
cos
g
I( ) sin +
ab
cos
2
1 2
g
J( ) cos
2a
2
(20)
sin
2a
(21)
where Y1 and Y2 are the characteristic impedances in the input and coupling guides, respectively, and B1 and B2 are the
backscattered fields in the two guides. The value of 1/ M
gives the turns ratio for the transformer in the lumped-circuit model.
Coupling-Slot Geometries
The seriesseries angle slot is a convenient element for
broadwall coupling between transverse waveguides because
the coupling is controlled by adjusting the slot angle without
changing the relative locations of the waveguides and a wide
range of coupling values is attainable. This slot and its equivalent circuit model are shown in Fig. 6. The coupling value
for this element at resonance, assuming a narrow rectangular
slot with zero thickness, is given by
a1 b1Y1 |B1 |2
a2 b2Y2 |B2 |2
M=
a2 b 2 g 2
a b g
1 1
I1 ( ) sin +
I2 ( ) sin +
g 1
2a1
g 2
2a2
J1 ( ) cos
J2 ( ) cos
(22)
WAVEGUIDE ANTENNAS
497
S(, ) = A(, )
N
Vn e jkz n cos
(24)
n=1
(a)
1
..
M1/2
Bs
(b)
Figure 6. Seriesseries angle slot coupling element and its lumpedelement circuit model.
M=
1 a2 b 2 g 1 g 2
4
a31 b1
1 (l/a2 )2
1 (2l/g 1 )2
2
498
WAVEGUIDE ANTENNAS
Y1
Yn
N
Gi
P rn
G
= 1
G0
G0 n=1 P r1
d
Figure 7. Waveguide linear array of offset longitudinal shunt-slot
radiators and its transmission line circuit model.
Vi
A
=
C
Ii
B Vi+1
D Ii+1
(25)
cos(d)
j
sin(d)
Z0
jZ0 sin(d)
cos(d)
(26)
The transmission matrices for the shunt element with admittance Y and series element with impedance Z are
1
Y
0
1
and
1
0
Z
1
(27)
Array Mutual-Coupling Effects. The simple circuit model design is useful for understanding the slot array or for determining a set of initial design values, but it is not accurate
enough for most applications. The assumption was made that
the slots in the array have the characteristics of an isolated
slot. In the array environment the slots are affected by the
presence of the other slots, and slots at the edge of the array
exhibit behavior that differs significantly from that of slots in
the center of the array. If neglected, these effects result in
an array where the input reflection coefficient and radiation
pattern depart significantly from the desired values. The
characteristics of the isolated slot were determined by using a
single source of excitation for the slot: the incident waveguide
mode. Multiple sources in the array environment excite the
slot: the incident field, the fields scattered in the waveguide
by the other slots in the guide, and the radiated fields from
all the other slots. The characteristics for each slot are influenced by all the other slots in the array to some extent, so
a slots configuration cannot be determined independently. A
series of simultaneous equations is required to perform the
array design. Elliott (17) has developed a series of design
equations that account for the internal and external mutual
coupling effects:
Ypa /G0
f p Vps Vn
f n Vns Vp
(30)
2 f n2
Yna
=
N
G0
2 f n2
Vms
+ j(10 /k)(k0b)(a/)3
gmn
Y /G0
Vs
m=1 n
or
(28)
f n (xn , ln ) =
xn
(/l) cos(10 ln )
sin
2
2
(/ln ) 10
a
gmn (xn , ln , xm , lm ) =
Yna /G0
N
Yi
Yn
=
Y0
Y
n=1 0
N
Zi
Zn
=
Z0
Z
n=1 0
Gn
Gm P rn
=
G0
G0 P rm
(29)
l m /2
l m /2
cos(m /lm )
cos(n /ln )
l n /2
l n /2
jk R
0
2 e
dn m (31)
+
k
0
2
n
R
2
WAVEGUIDE ANTENNAS
499
The Yna /G0 are the active slot admittances in the array environment, the Vsn, Vn are the slot and the waveguide mode voltages, and the gmn terms represent the external mutual-coupling effects between the slots. The set of equations is applied
iteratively along with the equation for the required input impedance to determine the slot offsets and lengths. These equations assume an ideal cosinusoidal distribution in a narrow
slot, and may not be accurate enough for some applications.
A more thorough analysis, where the field distribution in the
slots are unknowns which must be determined, provides a
more accurate model of the impact of element mutual coupling on array performance (26). The computational requirements for a rigorous analysis may limit the size of the array
for which it is applicable. A good approximation to the final
design should be found first by using an approximate method
to reduce the number of iterations required.
=
N
Yn /Y0 e j2N d
n=1
2+
N
Yn /Y0
N
Yn /Y0 e j2n d
n=1
e j2N d
n=1
N
j2n d
n=1
The array excitations also change away from the center frequency because the slot locations relative to the standing
wave in the guide are no longer optimum and the slot radiation characteristics also exhibit frequency variation. This
causes pattern degradation and main-beam distortion. The effects are modeled by using the circuit model to determine the
slot excitations and computing the far-field pattern generated
by these excitations.
Traveling-Wave (Nonresonant) Arrays
For the nonresonant array, the array distribution is generated by progressively coupling energy from the waveguide
mode as it travels down the guide. The element spacings are
chosen to be nonresonant, and the waveguide is terminated
with a matched load. The design process differs from that of
the standing-wave array because the reflections from the slots
do not add coherently at the input. The input reflection is
controlled by the destructive interference due to the nonresonant spacing, and the reflection coefficients of the individual
elements are kept small to minimize the reflection. The traveling-array input-match bandwidth does not decrease as the
array size increases, unlike that of the resonant array. The
resonant slot spacing case for a broadside beam is avoided
(33)
i =
N
n e j2n d
(34)
(35)
n=1
(32)
Yn /Y0 e
cos 0 = /g /2d
2G/G0
2 + G/G0
(36)
P rn
Gn /G0 n1
Gi
=
1
P1r
G1 /G0 i=1
G0
(37)
where the product term gives the decrease in the power along
the guide caused by radiated energy. The ratio of the power
remaining after the final element is absorbed in the load termination relative to the input power is given by
N
Pl
Gn
=
1
Pi
Go
n=1
(38)
500
WAVEGUIDE ANTENNAS
S(, ) = A(, )
N
M
(39)
m=1 n=1
S(, ) = A(, )
(40)
m=1 n=1
S(, ) = A(, )
M
m=1
Vm e jkmd x sin
N
(41)
n=1
For arrays with separable distributions, the desired lineararray distributions are selected, and then the products are
used to determine the individual element distributions. The
Figure 8. Waveguide planar array with offset longitudinal shuntslot radiators and seriesseries angle-slot coupling elements.
WAVEGUIDE ANTENNAS
Rn /R0 =
1 M(n)
Gmn /G0
Mn m=1
(42)
Zi /Z0 =
N
1 M(n)
Gmn /G0
Mn m=1
n=1
(43)
r
Pmn
r
P1n
(44)
Mn = M1
Gmn /G0
M(1)
m=1
m=1
M(1)
M(n)
Gm1 /G0
m=1
P rm1
(45)
P rmn
m=1
The values for the G1n and M1 must be chosen to create the
desired input impedance value. For compactness and improved bandwidth characteristics, it may be desirable to feed
the array at the center of the feed waveguide using an additional feed layer with a coupling slot. For a shuntseries coupling element where the waveguide terminated by a short at
one-quarter guide wavelength, the input admittance to the
array is given by
Yi =
Z
M0
(46)
501
502
WAVEGUIDE ANTENNAS
ductive-tuning irises at the tee input and the length and position of a septum on the back wall of the tee. The performance
of the reactive-feed network is sensitive to the loads placed
on the feed by the modules. To reduce this sensitivity and
provide broader band performance, matched power dividers
are used though they have higher loss and larger sizes.
Frequency and Power-Handling Effects. As the operating frequency of a resonant array moves away from the design frequency, the input match degrades and the pattern performance suffers. For arrays whose radiating guide and feed
waveguides are fed near or at their centers, the amplitude
and phase of elements at the ends of the guide diverge from
those near the center. For a single module this broadens the
main beam and increases the sidelobe level. For an array of
modules, this creates an error across the array with a period
equal to the modular spacing which creates error lobes in the
far-field pattern and reduces the array directivity. These effects are modeled by using a circuit model for the modules
and feed network.
The amount of power that the array is required to handle
also influences the design. The waveguide power-handling capability is reduced as the waveguide b dimension is decreased, so for high-power applications full height waveguide
should be used. For arrays which must operate at high altitudes, the waveguides are pressurized. Care should be taken
to avoid any obstacles which protrude into the guide in the b
dimension and mismatches should be minimized to reduce the
standing-wave effect. The amount of power is reduced as it
divides through the feed network and subarrays, so the
power-handing problem is less severe at the aperture.
Nonresonant Planar Arrays
The traveling-wave planar array is created by joining a set a
linear, traveling-wave arrays with a feed network. The sidewall edge-slot array is frequently chosen, particularly for applications where the array is scanned in the feed plane, because the linear arrays can be packed closely together. The
radiating arrays are designed similarly to the linear array,
but mutual coupling between the linear arrays must be taken
into account. It is generally preferable not to subdivide the
array and to end feed the radiating aperture to minimize the
power lost to the loads. The feed network can be traveling
wave or standing wave depending on the application. A linear
array of coupling slots provides a compact feed network for
many applications. Where beam squint in the feed plane is
not desired, a corporate feed network of waveguide power dividers is used.
WAVEGUIDES
503
28. N. A. Begovich, Frequency scanning, in R. C. Hansen (ed.), Microwave Scanning Antennas, Vol. 3, New York: Academic Press,
1964, Chap. 2.
4. K. S. Packard, The origin of waveguides: A case of multiple rediscovery, IEEE Trans. Microw. Theory Tech., MTT-32: 961969,
1984.
5. S. A. Schelkunoff, The impedance concept and its application to
problems of reflection, refraction, shielding, and power absorption, Bell Syst. Tech. J., 17: 1748, 1938.
6. S. A. Schelkunoff, A mathematical theory of linear arrays, Bell
Syst. Tech. J., 22: 80107, 1943.
7. C. L. Dolph, A current distribution for broadside arrays which
optimizes the relation between beamwidth and side lobe levels,
Proc. IRE, 34: 335348, 1946.
8. T. T. Taylor, Design of line source antennas for narrow beamwidth and low sidelobe, IRE Trans. Antennas Propag., AP-7: 15
28, 1955.
WINIFRED E. KUMMER
9. W. H. Watson, The Physical Principles of Wave Guide Transmission and Antenna Systems, Oxford, UK: Clarendon Press, 1947.
ATTENUATORS.
417
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
sources on a general surface is approximated by a line integral along the wires axis. This is discussed later.
Most wire modeling, even when approximating a closed
surface by a wire mesh or grid, is done with piecewise linear
(straight) segments. Because meshes used as approximations
to curvilinear surfaces cannot be made of planar square or
rectangular elements, this means that junctions of wires
meeting at variable angles are encountered. Although requirement (2) above is then almost certainly violated, the
thin-wire approximation remains accurate enough to produce
useful results. The added difficulty of including such higherorder effects as a circumferentially nonuniform longitudinal
current is evidently not commensurate with the improvement
that might be realized. Use of straight segments is also common in modeling such simple objects as circular loops, helices,
conical spirals, etc.
Guidelines have been developed over the years to help users of wire codes choose modeling parameters more likely to
lead to acceptable results. For example, when modeling circular loops, the centerline circumference of the polygonal approximation should equal the circumference of the actual loop
and a minimum of six wire segments must be used to model
the loop in the vicinity of its first resonance, that is, where
C 1. This relates to the fact that experience has shown that
nominally six wire segments per wavelength are needed to
achieve acceptable accuracy, when the sampling density of
the unknown current is wavelength-driven. When more segments are needed to properly represent the geometry of the
object being modeled, the required segment lengths are required to be much smaller, and the sampling density is geometry-driven. When approximating a solid surface with a wire
mesh as shown in Fig. 1, the maximum mesh openings are no
greater than 0.1 wavelength on a side. The radius of the mesh
wires also needs to be such that their total area is twice that
of the surface they represent, so that the wire area in each
direction of a local, orthogonal coordinate system approximates that of the surface being modeled (6), the equal-area
rule. A computed result showing that the front-to-back ratio
of a backscreen of parallel wires is maximized when the
Back
Front
2a
xe
103
0.5 f0
Front-to-back ratio
418
102
101
0.75 f0
f0
1.25 f0
1.5 f0
1
0.05 0.15
a/D
0.25
The purpose of the analytical treatment and its implementation to develop a numerical model, described later, is to assess
the performance characteristics of antennas of interest, of
which some of the more important are described here (1).
(2)
where sin is the place at which the driving voltage Vin is applied and Iin is the current there. Determining a realistic
value of Vin from a numerical model presents a problem because it is sensitive to the way the feed region is described
and it is difficult to replicate the physical arrangement with
good fidelity. Feedpoint-modeling errors affect both Gin and
Bin though uncertainty in Vin is sometimes resolved, as far as
its effect on Gin is concerned, by far-field integration. In a computer code like NEC that uses point matching where it is assumed that Vin Ein, with the length of the feed segment
(see section later on numerical implementation), this uncertainty is usually resolved by integrating Etan in the vicinity of
the feedpoint.
The Far-Field and Radiated Power. The power radiated by
an antenna is usually determined by integrating the far-field
power flow over a closed surface containing the antenna as
given by
Prad =
1
Re
2
1 2
E H ds Iin
Rrad
2
S
Re[(E H ) r ]
1
lim
(E
H
)
r
ds
Re
r
4r2
S
Re[(E
H
)
r
]
= 2r2
Prad
(1)
where Prad and Ploss are the radiated powers and Ploss arises
from the losses to which the antenna is subject. From a numerical and analytical perspective, the antennas admittance
Yin is normally the more accurately obtainable quantity, because the real conductance Gin is determined wholly by Pin,
whereas the reactive susceptance Bin is very sensitive to feedregion geometry (discussed later). The real and reactive components of the impedance, on the other hand, which is simply
the inverse of the admittance, is affected by whatever errors
arise in the susceptance. Furthermore, when there are no
losses, it is straightforward to obtain the conductance by simply integrating the far-field power flow, a quantity that is also
relatively insensitive to errors in the computed current distribution. By contrast, the susceptance requires an accurate solution of the antennas near fields, because it is related to the
stored power that they represent.
The admittance is defined as
GD (, ) =
(4a)
where Prad /(4r2) is equivalent to the power density of an isotropic antenna radiating the power Prad. The directivity D of
an antenna is defined as the maximum value of the directive
gain, or
D = max{GD (, )}
(4b)
and finally the gain is defined as in Eq. (4a) but including the
losses so that
G(, ) = 2r2
Re[(E H ) r ]
= GD (, )
Pin
(4c)
1
E (r) = n
n
4 j
(3a)
419
2
JS (r) G (r, r )d r
(5)
and
Hi (r) +
JS (r) = 2n
1
2
JS (r ) (r, r )d 2 r
(6)
where
(3b)
G (r, r ) = ( + k2 )g (R)
(r, r ) = I g (R)
(7)
Equation (1) is the EFIE and Eq. (2) is the MFIE, where the
is an outwardunknown surface-current density is Js(r), n
420
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Isotropic
radiator
(theoretical)
Small
dipole
Isotropic
Dipole
Polarization
Pattern
number
2.14
none
1.74
0.4
60
34
2.14
49
55
2.14
37
100
2.14
150
130
3.64
1.5
L
L > /2
Gain:
dB above
Thin
dipole
L
L = /2
L/D = 276
Thick
dipole
L
L = /2
L/D = 51
Cylindrical
dipole
L
L = /2
L/D = 10
Cylidrical
dipole
L
L =
L/D = 9.6
90
90
90
0 z 180
180
0 x
270
270
270
90
90
90
0 z 180
0 z 180
180
270
0 z 180
270
0 x
270
Extension of the previous infinite-medium IEs is straightforward for a perfectly conducting half-space, a common starting point for treating imperfectly conducting or electromagnetically penetrable grounds, an especially important
application of wire-antenna modeling. By convention, the
half-space occupies the region z 0. It can be deduced that
(1), because an object located above a PEC half-space is electromagnetically imaged or mirrored in it, the vertical components of the image electric currents flow in the same direction as the actual currents and the horizontal components are
421
Table 1. (Continued)
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Folded
dipole
Gain:
dB above
Isotropic
Dipole
Polarization
Pattern
number
L
d
L = /4
L/d = 13
6000
1.64
0.5
300
45
2.14
72
100
2.14
350
200
2.14
150
50
0.86
150
20
7.14
R
Folded
dipole
L
d
L = /2
L/d = 25.5
Biconical
x
40
L = /2
R
Biconical
x 60
L
L =
R
Turnstile
L
L = /2
L/d = 25.5
d
R
Folded
dipole over
reflecting
sheet
L = /2
L/d = 25.5
/8 above
sheet
y
L
d
R
90
90
90
0 z 180
180
270
270
90
90
z 180
0 x180
270
(8a)
GI (r, r ) = I R G (r, I R r )
I R = x x + y y z z
0 x
270
No signal
response
in this
plane
0 x180
(8b)
0 x
270
I (r, r ) = I R (r, I R r )
(8c)
(8d)
422
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Dipole
over small
ground plane
Isotropic
Dipole
Polarization
Pattern
number
z
D
L = /4
L/D = 53
l = 2
Gain:
dB above
28
40
2.14
150
45
2.14
50
16
2.14
72
200
2.14
50
300
2.14
20
25
14.14
12
l
R
Folded
unipole
over small
ground plane
L = /4
L/D = 53
l=2
L/d = 13
L
l
Coaxial
dipole
L
L = /4
L/D = 40
R
Biconical
coaxial
dipole
d
L
L = /2
d = /8
D = 3 /8
Disc-cone
or rod
disc-cone
L = /4
l=
D
R
z
L
l
R
Biconical
horn
L = 9 /2
D = 14
D
L
R
90
90
90
0 z 180
180
270
270
90
90
0 z 180
180
C(r )
s =
|C(r )|
0 x
270
gI (R ) =
r (x, y, z) = r (x, y, z)
0 x 180
270
270
with
0 x
90
270
e jk R
R
R = |r r |
0 x 180
(9)
Imperfectly Conducting Ground
Modified Image Theory or Reflection-Coefficient Approximation. When the half-space of interest is finitely conducting or
423
Table 1. (Continued)
Type
Configuration
Slot in large
ground
plane
L
L = /2
l/d = 29
Isotropic
Dipole
Polarization
Pattern
number
350
70
2.14
45
13
3.14
130
200
10.14
circ.
600
100
16.74
14.5
300
30
14.74
12.5
50
35
15.14
13
D = /
D/d = 36
Helical over
reflector
screen,
tube 6 long
coiled into
6 turns
/4 apart
Rhombic
y
600
z
L
Parabolic
with folded
dipole feed
( /2)
Gain:
dB above
Vertical
full-wave
loop
L = 9
l = 9 /2
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
z
R
D = 5 /2
Horn,
coaxial feed
y
l
L = 3
l = 3
z L
y
No signal
response
in this
plane
z
90
z 180
0x180
0 x
270
270
270
90
90
90
y
90
0 z 180
180
270
0 x180
270
0 x
270
(r r ) z
|(r r ) z|
(10)
424
1
k2s
4j s
s Ei(s)
1
4
s Ei(s)
1
4
s sjI(s)g(R)ds
C(r)
C(r)
C(r)
sI(s)g(R)ds
s sjI(s)g(R)
1 I(s)
s g(R) ds
j s
1
4j S
C(r)
I(s)
g(R)ds
s
s Ei(s)
C(r)
C(r)
I(s) g(R)s s
1
2
C(r)
Aejks Bejks
d( s)ejks
1
2( /)
g(R)( s) [( s)g(R)]
s
ds
s Ei(s)ejkss ds
C(r)
(u) 1, u 0; (u) 0, u 0
Magnetic Vector Potential Integral Equation for Straight Wires
(Hallens Integral Equation)
C(r)
1
2( /)
C(r)
s Ei(s)ejkss ds
ejkR
;
R
R [a(r)2 r r2]1/2;
k /( )1/2 is the wave number of the infinite medium in which the wire is located; the superscript i denotes an incident-field quantity;
and the denotes a unit-length vector.
a
After (1).
as
s Ei (s) =
and
1
4 j
C(r )
(r) = tan1
(11a)
where
RE =
RM =
(11b)
(y y )
(x x )
(11c)
Here RE and RM are the Fresnel plane-wave reflection coefficients for transverse-electric (TE) and transverse-magnetic
(TM) polarizations, respectively and the subscript IG signifies an imperfect ground. Also, (r, r) is the angle with respect to a normal to the interface and a straight line in the
vertical plane that joins s and s (known as the specular reflection point from optics), and (r) and (r) are the directional angles of the wire at r with respect to the x and z axes,
but these explicit dependencies are omitted for clarity. Note
90
45
2
(13)
[sin gHz cos gVz ]
+ k2 cos +
sz
2
g
ds
+ sin k2 sin cos( ) +
Ht
st
00
90
(a)
00
45
where the partial derivative with respect to t is in the direction of the horizontal projection of the wire at s. Also
(b)
Figure 3. The real component of the radial electric field of a horizontal, delta-functional current source in free space is shown as a
function of distance R and elevation angle near a half-space for
IG / 4, IG 0.001 S/m, and (b) IG / 16, IG 0 (17). The latter
exhibits the beating effect of waves in the upper and lower half-spaces
with two different wavelengths which the loss in the former eliminates.
IG
j IG = Er jEi
J ()e (z+z ) d
+ E 0
0
cos( ) E
gHz =
J ()e (z+z ) 2 d
k2
E + E 1
0
E
gVz = 2
J0 ()e (z+z )
d
(14)
0
E
E
= (x x )2 + (y y )2 + a2
= 2 k2
E = 2 E k2
gHt = 2
425
(12)
where the integrals are the field expansions mentioned previously, first derived by Sommerfeld (11,13). The quantity
Jn(x) is a Bessel function of order n and argument x, z and z
are the source and observation heights above the interface,
and the a in the expression for radial separation imposes
the minimum separation required by the thin-wire approximation. Note that these integrals are functions of only two
variables and z z.
The treatment thus far is limited to the one-sided problem, where the object being modeled is entirely contained on
one side of the interface, so that the source and observation
points are always in the same medium. Similar, but still more
complicated expressions result when the object(s) being modeled occupy both half-spaces, that is, the two-sided problem.
It is worth noting that the RCA has been extended to the twosided problem (14), but the latter primarily has been approached with the Sommerfeld theory or its equivalent. Ana-
30
20
B (mS)
G (mS)
20
10
10
0
0
4
k Lh
4
k Lh
Figure 4. These three results for the input admittance of a monopole antenna of
length h with air insulation b/a 2.5 in
water having IG / 80 j0.197 as a
function of the composite wave number
demonstrate the general validity of the
sheath model described above. The xs are
from NEC, the solid lines are independent
computations (21), and the solid circles
are measurements (22).
426
;;;
MODELING WIRE ANTENNAS
/4
/4
1.40
1.<180
replaces the Sommerfeld fields with a series of images in complex space that provide the interface-reflected fields. Another
alternative is to generalize the problem to include the tangential electric and magnetic fields on the interface as additional
unknowns in the model (16). Because the interface extends to
infinity, this is impractical because the number of unknowns
is commensurate with the area being modeled. However, in
terms of accounting for the interaction of an object with the
interface, only a limited area of the interface under the objects projection onto it measurably affects the objects currents. Thus, the number of interface unknowns is limited.
Furthermore, for a particular choice of ground parameters
and frequency, the interface part of the problem needs to be
solved only once and then continues to be reused for whatever
object is to be modeled for those specific conditions.
Approaches that fall in category (3) are of a generally different character and are described as employing a signal-processing philosophy. The basic motivation here is to compute
a minimum number of Sommerfeld-integral values, or their
equivalent, beyond which the field values needed in the IE
12.
10.
R ()
8.
N = 20
6.
N = 50
4.
N = 100
2.
g = 10 j100
0.
0.0
0.1
0.2
0.3
Screen radius/ 0
0.4
0.5
10.
Category (1) has long been pursued, and a variety of analytical approximations have been derived for various regions of
the parameter space for which Sommerfeld-integral values
are needed, as have been numerous numerical approaches.
Unfortunately, although some very useful results have been
obtained, individual approaches generally do not cover the parameter space encountered in many practical problems, so
this possibility has had only limited success. The Sommerfeld
integrals are one of the wave-type, or Fourier-type expansions, developed for the interface problem, of which a summary is found in (13).
As an alternative to wave expansions that lead to Sommerfeld-type integrals, one example of category (2) is the development of complex-image theory (15). This particular approach
8.
6.
X ()
N = 20
4.
N = 50
2.
N = 100
0.
2.
0.0
0.1
0.2
0.3
Screen radius/ 0
0.4
0.5
427
180
h/
120
160
0.25
100
Input resistance ()
Input resistance ()
110
0.30
90
0.35
80
0.45
70
140
h/
120
0.3
100
60
60
40
50
20
40 5
10
0 5
10
100
0.5
80
0.1
(a)
(b)
model are obtained from a reduced-order analytical approximation based on these values. A motivation for doing this is
demonstrated in Fig. 3 (17), where the real component in the
radial direction of the interfacially reflected field for a horizontal delta-function electric-current source is shown graphically as a function of the two applicable variables R
[2 (z z)2] and tan1[(z z)/ ]. The spatial variation of the fields is quite smooth, certainly less complex than
the analytical descriptions of Eq. (14) imply. Furthermore the
possibility is suggested that, if the fields are sampled over
a mesh, simple interpolation might provide accurate enough
values between these samples, so that no further rigorous
field computations are required in numerically modeling objects near an interface, as was shown feasible (18).
7
Ground screen
Re(Zin)()
d-layer T = 15,
= 0.01
240 radials
4
100
Figure 7. These plots of the input resistance of (a) a half-wave vertical dipole and
(b) a half-wave horizontal dipole as a
function of ground conductivity for various antenna heights compare the RCA
(solid line), the Sommerfeld theory (os),
and the compensation theorem (xs) for
the vertical antenna (10). The frequency
is 3.0 MHz, E 10, and the antenna radius is 5 104 wavelengths. The RCA
becomes exact for PEC and so is expected
to be most reliable for reasonably conductive grounds.
360 radials
3
0.1
10
100
d (meters)
Figure 8. These results for the input resistance of the SLT as a function of depth to a salt-water half-space show that between 240 and
360 radials are needed to limit the input-resistance change to less
than 10% (28). They are based on combining the RCA with Eqs. (19),
(21), and (24). Also, E,1r 15, 1 0.01, and E,2r 81, 1 4.
Figure 9. A hybrid of an integral equation and GTD model is used here to obtain
the input impedance of a monopole antenna at the center of an octagonal plate
(29). Agreement between the experimental results (open circles) and the computations (solid line) is within 10%.
64
36
60
32
56
28
52
24
Input resistance ()
Input resistance ()
428
48
44
40
34
20
16
12
8
32
28
24
0.2
0.4
0.6
d/
ES(s) represents the field scattered from the wire by the incident field, as follows (1,2):
1
s [Ei (s) + Es (s)] = s Ei (s)
I(s )G (s, s )ds
4 j C(r )
(15)
= ZL (s, s )I(s )
0.8
1.0
1.2
0.2
0.4
0.6
d/
0.8
1.0
1.2
tions, with the requirement that their effect is truly local and
that the load is itself nonradiating except for the current
flowing through it at its connection point si (2).
Loading Due to a Sheath. The effect of a sheathed wire is
handled by including an equivalent electric current Js radiating in an infinite medium as given by (2,20)
where s is the permittivity of the sheath relative to the infinite medium where the wire is located. So as not to destroy
the one-dimensional nature of the wire integral equation, the
sheath is assumed to be electrically thin, and the total sheath
field is assumed to be dominated by its radial component,
which is due to the wire charge, so that
where I(s) is the spatial derivative of the current at s. Because the incident electric field in the radial direction can be
neglected relative to that due to I(s), the equivalent sheath
current becomes
(16)
where
qi =
i i ai
Js (, s, ) = j (s 1)[Es (, s, ) + Ei (, s, )]
Es (s)
Js (, s, )
jI (s)
2 s
(s 1)
I (s)
2s
(18a)
(18b)
(18c)
jk2s (s 1)
b(si )
Ci Ln
2 s
a(si )
(18d)
6.00e4
G, B(mhos)
4.00e4
2.00e4
G One-segment
B Source
0.00e+0
and
0
20
40
60
80
100
Number of segments (unknowns)
(a)
10000
Resistance, reactance ()
Ex,IG = ZsurfHy,IG
G Constant-width
B source
4.00e4
5000
Resistance
Reactance
0
5000
Zsurf = IG /IG = / E
(19a)
where is the wave impedance of the upper medium. Then
the SIA is expressed as
2.00e4
6.00e4
429
(19b)
Ey,IG = ZsurfHx,IG
Zsurf Zscreen
Zsurf + Zscreen
(20)
100
200
Number of unknowns
(b)
Figure 10. The input impedance (b) and admittance (a) are shown
for an L 2, center-fed dipole antenna as a function of the number
of unknowns, where only the center segment is excited. The results
imply that the impedance is not well-converged whereas the admittance behavior suggests otherwise. When multiple segments are excited to keep the physical width of the source region a constant value
of L/11, the additional results of (b) are obtained. Its clear that physical and numerical effects influence the apparent convergence rate of
the numerical solution.
Zscreen,rad =
j
Ln
N
Na
(21a)
Zsurf + Zscreen
E = /Zsurf =
Zsurf Zscreen
j d
d
(22)
+ E 2 Ln 2a
=
d
j d
Ln
2
2a
430
Current
Charge
Figure 11. Current and charge snapshots at several time steps for a Guassian-pulse-excited dipole exhibit a variety
of important physics (31). Two outwardpropagating same-sign current pulses,
and opposite-sign charge pulses originate
from the source. A slight diminution of
the pulse amplitudes occurs because of radiation damping, and a further diminution occurs upon end reflection, a strong
source of radiation. Plots like these are remarkably insightful in revealing why antennas behave the way they do.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
4
3
2
1
(a)
4
3
2
1
(b)
4
3
2
1
2
1.5
0.5
Generator
Load
Distance, wavelengths measured from load
(c)
Figure 12. Comparison of current on a two-wire transmission line
obtained from the analytical solution (solid line) and from a thin-wire
integral equation (xs) for an (a) open circuit, (b) short circuit, and (c)
matched load (16). Although these results do not check the radiativepredictive properties of the numerical model, they do provide some
assurance that the model can handle a non-radiating problem.
103
431
B(plasma)
1000
X
+B(plasma)
100
B0
(free space)
Reactance ()
G1B(Mhos)
104
+X
105
G(plasma)
10
106
G0 103
(free space)
107
1.0
1.5
2.0
f(MHz)
2.5
3.0
3.5
Using Sommerfeld Theory to Model Ground Screens. Although intended primarily for modeling antennas near an interface, the Sommerfeld approach, as included in the NEC, is
also efficient enough to include the ground-screen wires for
some of the simpler designs. An example demonstrating this
application is presented in Fig. 6 (25).
Using the Compensation Theorem. An alternative approach
to including the effects of ground screens and the interface
itself on antenna impedance is a technique widely used before
computer solutions became practical, but still effective for
some applications. This is the compensation theorem (26),
whereby the input impedance in an antenna located near an
interface is estimated from a reference solution for a PEC
half-space, as
Zin,IG = Zin,PG +
Iin,PGIin,PG
A
(23b)
Figure 14. A comparison of the input impedance of a Vee-dipole antenna obtained from several different wire IE models shows that they
agree to an order of 30% except for the resistance at lower frequencies when decreases as f 2 (16).
200
Resonant frequency (MHz)
0.5
29.6 in.
190
180
1 in.
170
160
150
Experimental
Numerical
140
130
120
20
40
432
12
Below ground
10
Above ground
X 103 ()
R ()
Figure 16. Although the thin-wire approximation limits how close a wire can
approach an interface, by using a small
radius of 108 wavelengths in free space,
the 0.1 wavelength wire can be brought
much closer than its length to an interface
of relative permittivity 16 (17). The
above-ground results approach an asymptotic value, whereas those below ground
continue to change. The smoothness of the
behavior shown here does not prove the
validity of the model, but does show that
the model, based on the MBPE form of the
Sommerfeld approach, exhibits appropriate behavior.
6
4
3
Above ground
2
1
Below ground
0
101
102
103
h/ o
+
Zsurf = Zsurf E,1
E,2 +
(24)
104
105
106
106
Hybrid Models
Unfortunately, a numerical model based on a single formulation is not equally applicable to all problems. Even though a
wire code models surfaces approximated as grids, there are
more efficient and accurate ways to handle PEC objects. It is
equally true that not all parts of a given problem are suitable
for modeling with a single formulation. Wires attached to
solid PEC objects or located near inhomogeneous dielectric
bodies are simple examples. For this reason, it is desirable to
combine two (or more) approaches to develop what is called a
hybrid model to handle such problems.
Considering that there are four kinds of numerical models
presently used in electromagnetics (EM), based on integral
equations, differential equations, modal expansions and the
geometrical theory of diffraction (GTD), formulated in either
the frequency domain or time domain, hybrid models have
been developed using various combinations of these (see
(29,30) for two examples). An especially useful hybrid model
involving wires is one that combines a wire IE with the GTD,
an example of which is shown in Fig. 9. The benefit of this
kind of hybrid model is that only the unknowns for the cur-
6
5
4
3
2
1
0
1
2
3
4
5
6
0.8
E ( )/E(a)
1.0
0.6
0.4
0.2
0.5
1.5
2
2.5
3
3.5
4
4.5
Distance from source in wavelengths
5.5
Figure 17. The real (x) and imaginary (o) components of the current
on a conical spiral antenna obtained from a Pocklington IE compared
with a Hallen IE (solid and dashed lines) (36) show agreement to
within about 10% (16). The difficulty of generalizing the Hallen IE to
arbitrary geometries restricts its application.
11
/a
Figure 18. The normalized, near, radial electric field of a circular
loop antenna obtained from the electric-field IE (os) (16) agrees
within a few percent, on a normalized basis, with analytical results
(solid line) (37).
433
25
Frequency = 25 MHz
7.5 ft
Experimental
15
10
20
E-plane pattern
70.8 ft
10
0
164
Numerical
166
168
170
Frequency (MHz)
172
14
12
(25)
G(millimhos)
s s I(s , t )
s Ei (s, t) =
4 C(r )
R
t
c s R
I(s , t ) c
Q(s
+
,
t
)
ds
R2
s
R
t
I(s , t )
Q(s , t ) =
dt
s
10
8
6
4
2
0
8
6
B(millimhos)
10
20
H-plane pattern
174
Input impendance ()
20
4
2
0
2
4
0
4
6
Frequency (kL)
10
434
10
Resistance
600 ft
A
(Feed point)
14
70
ft
1470 ft
Figure 22. The computer model of the SLT antenna, containing a
total of 237 wire segments demarcated by the arrows (28). At the time
this work was done, a 237-segment model was state-of-the-art, but
now this problem is considered small for desk-top computers.
Resistance or reactance ()
0
Reactance
10
20
30
90
f (s ) =
Ns
ai f i (s )
(27a)
i=1
L(s, s )
Ns
ai f i (s ) =
Ns
i=1
(26b)
i=1
95
100
Frequency (kHz)
105
110
Figure 24. The modeled (os) SLT resistance results compare within
a few percent of the scale-model measurements (solid lines) while the
reactance values are shifted by about 2% in frequency (16) for 120
radials in the ground screen and E 15 and IG 2 102 S. Such
reactance shifts are fairly common in numerical modeling, and can
have a number of causes (34). Of course, besides limitations in the
numerical model, incomplete knowledge of the experimental conditions can also be a source of such differences.
where Ns is the number of spatial unknowns. Then upon sampling the operator Eq. (26b) with a set of testing or weight
functions
{w j (s)}, m = 1, . . ., M
R
10
X
17
16
15
14
13
12
11
10
9
(27c)
100
200
50
11
250
Xin (ohms)
i=1
Rin (ohms)
Input impedance ()
Ns
(27b)
150
100
50
Perfect ground
90
180
270
Number of 8 AWG radials
360
50
100
0.1
0.2
d/ 0
0.3
150
0.1
0.2
0.3
d/ 0
Figure 25. A numerical model like NEC makes it possible to evaluate the influence of parameters such as the length of a ground stake
against which a monopole antenna is driven, as illustrated here (24),
computed using the rigorous Sommerfeld approach for a wire radius
of 2.5 106 (solid line), 2.5 105 (dotted line), and 2.5 104
(dashed line), respectively, for a ground of E 16 j16. Results like
this make it feasible to determine what kind of ground stake is
needed to stabilize the input impedance.
Monopole
Feedpoint
L
a
2 Wire
1 Wire
3 Wire
(a) Elevation pattern
Er =
Dipole
(b) Azimuth pattern
Ns
Yij E ij
(b)
Dipole
(c)
Figure 26. Sparse ground screens like that shown here (a) cannot be
modeled using the simple ground-screen formulas, but can be handled
using the interpolated Sommerfeld approach (38), to obtain the elevation (b) and azimuth (c) patterns, with the number of ground wires
as a parameter. The frequency is 10 MHz and the ground has E 4
and IG 103 S.
Zji ai = b j , j = 1, . . ., M
(30b)
(31b)
where i is the length of segment i and (s sj) is a deltafunction, thus producing point sampling of the tangential
electric field. Of the 3Ns unknowns in Eq. (31), 2Ns are eliminated by enforcing current and current-slope (charge) conti-
(28a)
4
where
IZ Z/Z %
i=1
and
(30a)
j=1
1 Wire
Ns
Eio
where the first sum yields a solution for the current distribution and the second sums the field at a specified observation
point caused by that current. Also
(a)
3 Wire
Ns
i=1
Continuous
wire
2 Wire
435
ai =
M
Yij b j , i = 1, . . ., Ns
(29)
j=1
ZA =BA = YB
10
Separation (ft)
Figure 27. Comparison of the TCA (dashed line) with the Sommerfeld theory (solid line) demonstrates the formers potential applicability to subsurface probing (14) for a ground of E 9 and IG 103 S
and with both wires 0.5 wavelengths long and for a frequency of 10
MHz. The goal here is to detect a buried object from its effect on the
impedance of an above-ground probing antenna whose effect is plotted as a normalized difference between its value with and without
the object.
436
9
8
(33b)
7
6
Nt
Ns
i=1 j=1
3
2
Crossed
Parallel
1
0
10
20
30
40
50
(deg)
60
70
80
90
Figure 28. Here the TCA is used to determine the variation in the
input impedance of an above-ground, horizontal antenna (L 0.457
wavelengths) as it rotates about a vertical axis above a horizontal,
buried wire (L 0.152 wavelengths) (14) whose center is on the same
vertical axis with a ground of E 9 and IG 0 for a frequency of 1
MHz. When the wires are orthogonal, there is no interaction while
the maximum effect is produced when they are parallel.
(34a)
(34b)
= 0 (Horizontal)
;;;;
30
(32a)
60
F(s , t ) =
Nt
Ns
i=1 j=1
(33a)
90
30 ft
(32b)
=0
r= 9
50 ft Rotated antenna
Figure 29. The TCA is used here to determine the above-ground radiation pattern of a dipole antenna as it rotates in a vertical plane
(14). When the antenna is vertical or horizontal, the above-ground
patterns are symmetric about a surface normal, but the patterns are
otherwise slightly skewed.
3
2
1
0
1
2
0
10
20
Time (ns)
437
0.2
0.1
0.1
30
10
(a)
20
Time (ns)
30
(b)
(35)
i =1
1
0.4
Resistance (k)
0.8
Resistance (k)
Experiment
Calculated
Experiment
Calculated
0.6
0.4
0.2
0.2
0.2
2
1
Frequency (GHz)
(a)
0.4
1
2
Frequency (GHz)
(b)
Figure 31. This comparison of the measured and computed, using TWTD, input
impedance of a Gaussian-pulse excited
antenna shows the two results are graphically indistinguishable for about the first
8 resonances (31), providing mutual validation, as well as demonstrating the
broadband capability of a time-domain
computation and measurement.
438
;;
Current (mA)
8
12
Time (ns)
16
20
439
0.12
Radiated field R V
1.8
1.2
0.6
0.06
0.0
0.06
0.0
0.2
4
Time (ns)
0.12
(a)
4
Time (ns)
(b)
E R magnitude
Vs
Hz
6.0
4.5
3.0
1.5
0.0
3
1
2
Frequency (GHz)
(c)
15
Relative magnitude
10
2
0
0.4
Time ( s)
(a)
0.8
1.2
20
40
60
Frequency (MHz)
(b)
80
440
BIBLIOGRAPHY
1. A. J. Poggio and E. K. Miller, Low-frequency analytical and numerical methods for antennas. In Y. T. Lo and S. W. Lee, eds.,
EDMUND K. MILLER
441
379
380
Apparent direction
to satellite
Real direction
to satellite
1
dn
=
r
dh
Horizontal
a (km)
(2)
1 1
1
dn
1
1
= +
=
=
a
r
a dh
ka
Re
Center of earth
(a)
(3)
Here Re is the effective radius of the earth given by the product ka, where k is the effective earth radius factor. The effective radius of the earth is often used in ray-tracing diagrams
of the vertical profiles of terrestrial radio systems, because it
allows for the average ray bending along the path [see Fig.
1(b,c)]. Ray bending on satellite-to-ground paths may cause
signal loss due to the spreading of the antenna beam. This
effect, and that of elevation-angle-of-arrival variations, are
negligible at elevation angles above about 3 and will only be
significant for large antennas (diameter 100 wavelengths).
The atmosphere has a dry and a wet refractive index component, the former due to the gases in the atmosphere (principally nitrogen and oxygen), and the latter due to the water
vapor that is taken up into the atmosphere. Combining the
two components together into a composite refractive index n
gives
a (km)
C
(b)
ka (km)
n1=
77.6
e
P + 4810
106
T
T
(4)
where
C
(c)
Figure 1. (a) Apparent and real directions to an earth satellite; illustration of methods of ray tracing: (b) real radius and a curved ray
path, and (c) effective radius and a geometric ray path.
this reason, an electromagnetic signal (a ray) leaving the surface of the earth at an elevation angle of 90 will bend away
from the normal to the surface [see Fig. 1(a)]. The curvature
will be contained in one plane, the projection of which onto
the surface of the earth is the azimuth direction. If r is the
radius of curvature of the ray (2), then
cos dn
1
=
r
n dh
(1)
where
(5)
bt
t+c
where
H relative humidity (%)
t temperature (C)
es saturation vapor pressure (hPa) at the temperature t
and the coefficients a, b, and c are as follows:
(6)
For water:
a = 6.1121
b = 17.502
c = 240.97
For ice:
a = 6.1115
b = 22.452
c = 272.55
The vapor pressure e is related to the water vapor density
(g/m3) by (3)
e=
T
216.7
(hPa)
(7)
(8)
(9)
where
h height above mean sea level (km)
h0 reference altitude 7.35 km
N0 average value of atmospheric refractivity at mean sea
level 315
Rewriting Eq. (9) in terms of the refractivity, we have
N(h) = N0 exp(h/h0 )
157
0
40
157
200
1
3.65
Duct Thickness
(m)
0.1
0.14
0.22
10
20
50
(11)
(10)
381
(mrad)
(12)
382
later than expected). These are the range delay d due to dry
air and the range delay w due to the moisture in the air.
Typical values (6) are given in the table below for a path that
exits the atmosphere to a satellite where the refractive index
of the atmosphere is assumed to be unity. The range error
values have been increased by 2% from those in Ref. 5 to
allow for the additional atmosphere encountered from the 80
km height used in Ref. 5 for the target. Regression analyses
(7) have shown that these typical range errors can be reduced
in practice by 25% by incorporating site latitude, height above
sea level, and time of day within the calculation procedure
Initial Elevation
Angle (deg)
Elevation Angle
Error (mdeg)
0
5
50
595
159
13.4
106
25.5
3.2
;;;;;
Heavy thunderstorm rain
10
Performance
threshold (108)
log (BER)
7
6
5
4
Availability
threshold (106)
Range of degraded
performance
(i.e. < 108
but > 106 BER)
3
2
1
0
0.001
0.01
0.1
1
Annual time percentage
10
100
Step 9. Calculate the scintillation fade depth for the time percentage p by
A s ( p) = a( p)
dB
5/6 1/2
Step 8. Calculate the time percentage factor a(p) from the time
percentage, p, of concern in the range 0.01 p 50,
a( p) = 0.061(log10 p) 3 + 0.072(log10 p) 2
1.71 log10 p + 3.0
dB
The above procedure has been found to predict well at frequencies at least up to 30 GHz, although it tends to underpredict slightly at all frequencies in the hot, humid summer
months. At higher frequencies (30 GHz) it is expected that
the same f 7/12 scaling law will apply as shown in step 7 above,
although this anticipated result has not been tested experimentally. The fading portion of the tropospheric scintillation,
As, is not usually significant for earth-to-space paths at elevation angles above 20 for the frequency bands currently in
operation (4 GHz to 30 GHz). As the elevation angle decreases, or the frequency band increases, tropospheric scintillation fading becomes more significant. For example, at 11
GHz, As is less than 0.5 dB for 0.01% of an average year at
elevation angles on the order of 30. For the same frequency,
As approaches 3 dB at elevation angles of 3.2 (11). These values of As are for nonraining periods and so will affect the performance of the system.
GASEOUS ABSORPTION
When a molecule or atom is nonsymmetrical in its structure,
it will have a preferred orientation if placed in an electric
field. Such molecules or atoms are said to be polar. Polar molecules or atoms cause far more loss to a radio signal than
nonpolar molecules or atoms, since the application of an external field causes the polar molecules or atoms to reorient
themselves. The reorientation (or relaxation) of the dipoles
will remove energy from the applied field and cause heating
of the medium. This is the principle of a microwave oven. The
major constituents of the troposphere, oxygen and nitrogen,
are electrically nonpolar and so no absorption due to electric
dipole resonance will take place. Oxygen however, is a paramagnetic molecule with a permanent magnetic moment
which causes resonant absorption at particular frequencies.
Water and water vapor, which both contain two oxygen
atoms, are therefore polar molecules and so absorption occurs
due to resonance at critical frequencies.
Oxygen and Water Vapor Resonance Lines
The specific gaseous attenuation (dB/km) is given by
= o + w
with
383
(dB/km)
(13)
(dB)
(14)
For earthspace paths, the situation is not so straightforward, since the density and make up of the atmosphere
change rapidly with height. The value of L used must be able
384
(dB)
(15)
where ho is the equivalent height of the dry gaseous absorption and hw is the equivalent height of the water vapor absorption. Typical values of ho and hw are 6 km and 2.1 km
(during rainy conditions), respectively. For elevation angles
between 10 and 90, a cosecant law gives the total gaseous
attenuation A, namely
A=
Az
sin
(dB)
(16)
105
104
103
102
10
101
Figure 3. Zenith attenuation due to atmospheric gases (from Fig. 3 of Ref. 14).
Curve A: standard atmosphere (7.5
g/m3); curve B: dry atmosphere. ITU.
Reproduced with permission.
102
0
100
200
300
400
500
600
Frequency (GHz)
700
800
900
1000
Step 1. Calculate the effective rain height, hR, for the latitude
of the station, :
49.5 GHz
5 0.075( 23)
5
5
h R (km) =
5 + 0.1( + 21)
40
39.6 GHz
30
18.7 GHz
0
0
71
20
10
12 GHz
0
0.001
385
0.01
0.1
1
10
Percentage of a year attenuation is exceeded
for
for
for
for
for
100
L s = (h R h S )/(sin )
Figure 4. Average annual cumulative statistics of total path attenuation at an elevation angle of about 38 in Europe (some data extracted from Ref. 16).
km
Step 3. Calculate the horizontal projection, LG, of the slantpath length from:
(dB/km)
L G = L S cos
r 0.01 = 1/(1 + L G /L 0 )
where L0 35 exp (0.015 R0.01) and, for R0.01 100 mm/h, use
the value 100 mm/h in place of R0.01.
Step 6. Obtain the specific attenuation, R, using the frequency
dependent regression coefficients k and [Ref. 18] by using
where k and are regression coefficients (18) that take account of absorption and scattering.
R = k(R 0.01 )
Prediction of Attenuation
km
(17)
km
dB/km
Step 7. The predicted attenuation exceeded for 0.01% of an average year is obtained from
A 0.01 = R L S r 0.01
dB
Step 8. The estimated attenuation to be exceeded for other percentages of an average year, in the range 0.001% to 1%, is de-
Earthspace
path
Rain height
hR
Region of liquid
precipitation
LS
hR-hS
hS
LG
386
p)
Note that the variability of the weather from year to year will
cause this prediction method to vary similarly. Average accuracy
better than about 35% should not be expected for any given year.
for 8 f 35 GHz
DEPOLARIZATION
Step 2. Calculate the rain attenuation dependent term
The polarization sense of a wave usually refers to the preferred orientation of the electric vector. A perfectly polarized
wave will have no component of its electric field in the orthogonal sense. There are only two states for a perfectly polarized
wave: linear and circular. For linear polarization, it is common to use vertical and horizontal as the two orthogonal (linear) reference axes. For circular polarization, right-hand circular polarization (RHCP) and left-hand circular polarization
(LHCP) are the two orthogonal (circular) reference senses.
Depolarization from one reference polarization sense to the
other occurs when the signal passes through an anisotropic
medium and the polarization sense of the signal is not aligned
with the principal axes of the medium. Rain can be such a
medium.
Raindrops exist in a dynamic medium: the atmosphere.
When the drops are very small, surface tension forces (which
strive to keep the drop perfectly spherical) far exceed the
aerodynamic forces on the raindrop (which seek to shape the
drop to conform with the external pressures). Under these
conditions, the raindrops remain spherical and the rain medium appears symmetrical to the signal. No signal depolarization therefore results. As the drops become larger, however,
through coalescence with other drops on collision in the rain
medium, the surface tension forces can no longer keep the
drops spherical. The drops distort into elliptical shapes and
so exhibit preferred orientation directions along the major
and minor axes of the ellipsoid. Differential attenuation and
phase shift will occur on component signals propagating along
these axes and will result in signal depolarization. The degree
of depolarization measured is called the cross-polarization
discrimination (XPD) and is defined as
XPD = 20 log10
Eco
Ecross
(dB)
(18)
C A = V ( f ) log A p
where
V ( f ) = 12.8 f 0.19
V ( f ) = 22.6
for 8 f 20 GHz
for 20 < f 35 GHz
for 60
dB
dB
dB
The component for ice depolarization in the above formulation, Cice, is known to underestimate the degree of ice crystal
depolarization encountered on low-elevation-angle paths.
Depolarization can be system-limiting in the lower frequency bands, where rain attenuation is not significant (e.g.,
in the C band, which is the 6/4 GHz uplinkdownlink pair
for many satellite systems). As the frequency goes up, however, the amount of depolarization observed per decibel of at-
387
388
RELATIONAL DATABASES
JEREMY E. ALLNUTT
Virginia Tech/Northern Virginia
Center
128
RADIO NOISE
RADIO NOISE
Radio noise can be defined as unwanted and unavoidable electromagnetic fluctuations that tend to obscure the information
content of a desired radio signal. In a radio receiver, the noise
disturbances are fluctuating voltages and currents that alter
the desired signals original frequency, amplitude, or phase in
some unpredictable, unwanted manner. In early radios, much
of the noise stemmed from the quality of the equipment; for
example, poor contacts and unreliable connections, sensitivity
to mechanical vibrations, and unstable detectors and amplifiers all contributed to the degradation of the radio signal.
Over the years, there was steady progress in improving transmitters and receivers and greatly reducing or completely
eliminating these and other noise sources. It was soon realized, however, that all sources of noise could not be totally
eliminated and that a fundamental residual noise remained
as a lower limit. This article focuses on such noise emanating
from both natural sources and human activities that affect
the performance of narrowband radio frequency (RF) communications systems.
Typically radio noise is divided into two components
internal or receiver noise (that is, intrinsic noise generated by
the receiving system components) and external or environmental radio noise (noise collected by the receiving system
antenna). Receiver noise is largely due to natural processes
such as thermal noise, arising from the motion of thermally
agitated free electrons in resistors, and shot noise, resulting
from current fluctuations superimposed on the steady current
flow in both vacuum tubes and solid-state devices. Environmental noise is an intrinsic part of a wireless communications
channel and consists of natural radio noise (i.e., noise originating from natural sources such as atmospheric noise and
sky noise), and unintentional RF emissions due to human activity, which are commonly referred to as man-made noise.
Atmospheric noise is largely due to lightning and related phenomena, such as precipitation static. Sky noise is a generic
term used to describe noise from a variety of terrestrial and
extraterrestrial sources, such as cosmic noise and absorption
of radio waves in the earths atmosphere due to rain, water
vapor, and oxygen. Intentional RF emissions that result in
adjacent- and co-channel radio interference, including intermodulation and jamming, are not covered in this article.
While such noise cannot be completely eliminated, understanding the character of radio noise allows engineers and
system designers to evaluate radio system performance and
devise means to lessen the adverse effects. The most important design parameter used to characterize noise performance
is the predetection signal-to-noise ratio (SNR), and hence,
noise power is generally the most significant parameter in relating the interference potential of noise to system performance. Noise power calculations must include noise generated in all components of the receiving system as well as the
radio channel. This is usually accomplished by determining a
single parameter, the operating system noise factor, which is
essentially the ratio of the receiving system predetection
noise power referenced to the terminals of an equivalent lossless receiving antenna and the noise power available from a
resistor at standard temperature (288 K).
In this article, important sources of receiver noise are described and the concept of receiver noise factor is developed.
The noise factor concept is then extended to cover the entire
radio receiving system, including the receiving antenna and
environmental noise. This is followed by a description of important environmental noise sources.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
RADIO NOISE
(1)
dws
d
(2)
Solid-state shot noise sources such as noise diodes are commonly used in noise measurements. To achieve the required
precision in Te, these devices are calibrated against a primary
standard. Noise diodes are commonly characterized in terms
of excess noise ratio defined as
NR = 10 log10
T
T0
(dB)
(3)
Thermal noise generated within or passing through a bandlimited receiver circuit is colored and perhaps amplified.
Since thermal noise power is independent of frequency over
the bandwidth of most narrowband RF communication systems, it is useful to define a noise-equivalent bandwidth b0.
When a resistor is attached to the input terminals of a linear
receiver with gain g(), the total noise power available to the
receiver wa in the system band can be expressed in terms of
the noise equivalent bandwidth as follows:
wa = kT
0
g()
d = kTb0
g0
(4)
129
(5)
(6)
(7)
130
RADIO NOISE
In radio engineering it is desirable to have a simple, yet unambiguous method of characterizing the noise properties of a
radio receiver. For this purpose, the concept of noise factor
was developed and defined. There are a variety of ways to
define noise factor. The basic idea is to relate the available
noise power at the input of an active or passive two-port linear network to the noise power available at its output terminals. Assume that a CW signal generator is connected and
matched to the input terminals of a receiver. Let wg be the
available signal generator power, wl be the signal power available at the load, and w0 be the available noise power at the
load; the International Radio Consultative Committee (CCIR)
(6) definition of noise factor is given as follows: The noise factor f r of a radio receiver is defined to be the ratio of the available CW signal-to-reference noise power wg /kT0b0 at the terminals of the signal generator to the corresponding total
signal-to-noise power ratio wl /wo available to the load of the
linear portion of the receiver with the CW signal tuned to the
maximum response of the receiver bandpass characteristic
and with the signal generator output impedance at the reference temperature T0:
wg /kT0 b0
wg wo
=
wl /wo
wl kT0 b0
(8)
fr
wo
go kT0 b0
(9)
dw0
d
f s () =
g()kT0
(10)
Communications receivers use transmission lines, attenuators, and other devices that can be generally classified as passive two-port linear networks. It is important to consider the
effects of such devices operating in tandem with the actual
receiver to characterize properly the noise performance of the
receiving system.
The spot noise factor of a linear two-port network is obtained by calculating the power density ratio given in Eq. (10)
when a resistor at reference temperature T0 is connected to
the input terminals of the network. If the network is at the
ambient temperature T0, the available output noise power
density is just kT0. If the network temperature is Tn T0, a
correction arising from the temperature difference T0 Tn
must be included; hence
dw0
= kTn + gk(T0 Tn )
d
(12)
where g is the available gain. From Eq. (10), the spot noise
factor is
fs = 1 +
Tn
Tn
(1/g 1) = 1 +
( 1)
T0
T0
(13)
where 1/g is the available loss factor. Note that the spot
noise factor depends on the input resistance used to determine the available gain (or loss).
Noise Factor for Cascaded Linear Networks
The noise factor for a linear receiving system can be calculated from the noise factors for its individual components. Understanding the relationship between the system noise factor
and the noise factors for various components is important in
system design since it shows which components contribute
most significantly to the noise factor.
Assume that two networks are connected together: network a with spot noise factor f a and gain ga, and network b
with spot noise factor f b and gain gb. Further assume that a
resistor is connected to the input of network a and the entire
system is at temperature T0. The power density available at
the output of network b is equal to the sum of the power density available at the input of network b scaled by gb and the
available noise power density due only to network b or
dwo
= f a ga gb kT0 + ( f b 1) gb kT0
d
(14)
Using Eq. (10), the spot noise factor for two cascaded networks is
f ab = f a +
fb 1
ga
(15)
RADIO NOISE
where f a, f c, and f t are the spot noise factors; and ga, gc, and
gt are the gains of the antenna, antenna circuit, and transmission line, respectively; and f r is the spot noise factor of the receiver.
The external noise power in the band d that is available
at the terminals of the loss-free receiving antenna can be expressed as kTa d, where Ta is the noise temperature of the
radiation resistance of the receiving antenna at the center frequency of the receiver. The antenna noise factor f a is equal to
the ratio Ta /T0. Since the antenna circuit and transmission
line are passive two-port networks, the spot noise factor can
be expressed as
f = Ta /T0 + (c 1)Tc /T0 + c (t 1)Tt /T0 + c t ( f r 1)
(18)
where the gains have been replaced by the corresponding loss
factors. Strictly speaking, this result is meaningful for narrowband systems where the noise factors do not vary significantly over the operating frequency range. It should be understood that all of the spot noise factors and loss factors are
determined with generator impedances that are the same as
those in the actual receiving system. Also, the formula for
cascaded networks is applicable regardless of the degree of
match or mismatch between the output impedance of one network and the input impedance of the following network. The
magnitudes of these mismatch losses do influence the values
of the spot noise factors and loss factors and will thus affect
the noise factor.
The quantity of interest for finite bandwidth systems is the
weighted average spot noise factor, which is often referred to
as the operating noise factor f op (6). The operating noise factor
can be expressed in terms of the weighted average noise factors of the receiving system components as follows:
f op = f a + f ct 1 + (br /b)( gr /g0 )( f r 1)
(19)
fa =
fc 1
f 1
fr 1
+ t
+
ga
ga gc
ga gc gt
131
(17)
Ta () g() d
T0
g() d
(20)
Antenna
circuit
Transmission
line
Receiver
Output
(a)
(b)
(c)
(d)
(e)
la = 1
lc
lt
Ta
Tc
Tt
fa = Ta/T0
fc = 1 + (lc 1)(Tc/T0)
ft = 1 + (lt 1)(Tt/T0)
fr
132
RADIO NOISE
f ct is the weighted average combined noise factor of the antenna circuit and transmission line network,
f ct () g() d
f ct = 0
(21)
g() d
0
b is the equivalent bandwidth of the system, br is the equivalent bandwidth of the receiver, g0 is the nominal gain of the
system, gr is the maximum gain of the receiver, and f r is the
weighted average receiver noise factor, as given in Eq. (11)
and using the receiver gain function.
To apply the foregoing results, noise factors for the receiver, related electronic components, and an equivalent lossless antenna must be determined. Using the definitions given
in the previous sections, measurement methods can be readily
devised and implemented to calculate accurately noise factors
for receivers and other electronic components (see, for example, Ref. 9). The determination of the antenna noise factor is,
in general, much more difficult. For RF communication systems, the antenna noise factor is often a nonstationary random process that varies with time, frequency, geographical
location, and receiving antenna characteristics. As a consequence, researchers have spent many years measuring and
compiling statistics that can be used to estimate the antenna
noise factor for various environmental noise sources. In the
case of man-made noise, not only is there a fast time dependence over fractions of an hour to days, but there is also a
relatively slow time dependence based on advances in technology. Significant changes in the background noise for a particular environment may occur within a few years, and as a
consequence published data can become outdated. Antenna
noise factors based on published noise statistics are described
in the following sections.
So far in this article, lowercase letters have been used to
represent noise factors, gains, and bandwidths. In noise analysis, it is common to express these quantities as decibels and
to use uppercase letters to denote that the quantity is in decibels. Also, when given in decibels, the noise factor is usually
referred to as noise figure; for example, the antenna noise
figure is
Fa = 10 log10 f a (dB)
(22)
2h 3
c2 (eh /kT 1)
(23)
(24)
= 2
(25)
where is the free space wavelength. As in the case of thermal noise in a resistor, the radiation power is directly proportional to temperature.
The power received by an antenna from an incremental
blackbody of area dA at temperature T in bandwidth d is
dw =
1 2
kT () d d
() d d =
2 4
4
(26)
where d is the element of solid angle subtended at the receiver by dA and () is the directive gain of the antenna.
Note that the antenna only collects half the power since only
a single polarization is received. The total power collected
when the blackbody radiation at temperature T is received
from all directions is
(W/m2/Hz/sr)
w = kT d
1
4
() d = kT d
(27)
RADIO NOISE
260
240
A:
B:
C:
D:
E:
F:
A
B
220
G:
200
H:
I:
180
J:
K:
LQ:
M:
N:
160
140
120
100
2.9 1026
2.9 1023
2.9 1020
2.9 1017
2.9 1014
2.9 1011
80
LQ
2.9 108
60
40
LQ
2.9 105
0
E
20
10
100 1 kHz 10
100 1 MHz 10
Frequency
H
M
290
N
J
100 1 GHz 10
0.29
100 1 THz
120
Atmospheric
100
Fam (dB above kT0b)
20
1 Hz
2.9 1029
280
133
Night
80
60
Day
Business
Residential
Rural
40
20
0
0.1
Quiet rural
Galactic
1
10
Frequency (MHz)
100
1000
RADIO NOISE
40
2.9 106
30
2.9 105
0
2.9 104
20
A
2.9 103
Fa (dB)
10
C
B
10
F
20
Sun
T ()p() d
= 0 Ts
2.9
2.9 101
30
40
108
109
(1 GHz)
1010
2.9 102
1011
Frequency (Hz)
2.9 10
E (90)
Curves LD, LQ, F, H, and M in Fig. 2 all refer to very narrowbeam antennas pointing directly at the source. Noise from
such sources (sun, atmospheric gasses, the earths surface)
are also expressed in terms of brightness temperature. These
curves can be used to calculate the antenna temperature of a
particular receiving antenna by integrating Eq. (25) in terms
of temperature over the region occupied by the noise source:
Ta = 0
T ()p() d
(28)
4 sources
Ta =
2.9 102
E (0)
ta (K)
134
2
1440
(29)
RADIO NOISE
140
130
120
110
100
90
80
70
60
50
Aiya (1955)
Aiya and Sonde (1963)
Hallgren and MacDonald (1962)
Schafer and Goodall (1939)
Japan URSI (1963)
40
30
20
10
Ep
0
10
Ep
20
30
1
Distance
Horner
and Bradley
(1941)
BW BW < 10 kHz
BW BW > 10 kHz
40
2
1 kHz
4 68
2
10 kHz
4 68
2
100 kHz
135
4 68
2 4 68
2
1 MHz
10 MHz
Frequency
4 68
2
100 MHz
4 68
2
1 GHz
4 68
10 GHz
Figure 5. Lightning emission peak field strength, 1 mile distant. (Reprinted from p. 369 of Ref.
18, by permission, 1982 IEEE.)
136
90
RADIO NOISE
60
75
90
105 120 135 150 165 180 165 150 135 120 105
90
75
60
45
30
15
15
45
60
90
40
35
80
30
50
55
80
70
70
60
60
60
60
50
50
45
55
40
40
70
30
30
60
20
20
10
90
95
85
10
75
85
90
0
10
8580
55
90
75
7085
10
80
20
20
30
30
75
75
40
60
55
50
70
65
40
70
50
60
60
50
45
70
70
80
80
90
40
60
75
90
105 120 135 150 165 180 165 150 135 120 105
90
75
60
45
30
15
15
30
45
60
90
Figure 6. Expected values of atmospheric radio noise, Fam (dB above kT0b at 1 MHz) (Winter
00000400 LT) (12).
ments of man-made radio noise. They devised methods for estimating the noise power and noise amplitude statistics that
are important in the design of radio systems. These methods
are described in the CCIR Reports (13) and have been widely
used by industry. Figure 3 summarizes these results in terms
of the median antenna noise figure Fam. As with atmospheric
noise, man-made noise is both nonstationary and nonGaussian and is a significant source of radio noise for frequencies below a few hundred megahertz. The antenna noise
figure Fa varies both in time and location. The noise level depends on the type and extent of human activities, which are
conveniently classified into four man-made noise environments (13) described in Table 1.
The within-the-hour time variability of Fa is commonly described by two log-normal distributions (21), as described previously for atmospheric noise. Values of Du and Dl are given
in CCIR Report 258 as a function of frequency and environment. More recently, Spaulding and Stewart (21) have analyzed the data used to obtain these decile values and have
found that it is appropriate to use the values Du 9.7 dB
and Dl 7 dB, independent of environmental category and
1 % locations
2L erfc
50
(30)
137
20
0.01
20
40
60
80
100
120
140
160
180
0.2 0.3
50
7 10
20 30
50 70 100
0
0.01
10
12
14
16
18
20
Du
Fam
Vdm
0.5 0.7 1
De
Dt
Du
Frequency (MHz)
0.2 0.3
Vd
Figure 7. (a) Variation of radio noise with frequency (winter; 0000400 LT) (12).
Expected values of atmospheric noise. Expected values of galactic noise. (b) Data
on noise variability and character (winter; 0000400 LT) (12). Fam: Standard deviation of values
of Fam. Du: Ratio of upper decile to median value, Fam. Du: Standard deviation of values of Du.
Dl: Ratio of median value, Fam, to lower decile. Dl: Standard deviation of values of Dl. Vdm: Expected value of median deviation of average voltage. The values shown are for a bandwidth of
200 Hz. Vd: Standard deviation of Vd (12).
Frequency (MHz)
0.5 0.7 1
20
30
40
60
70
80
90
100
(dB)
7 10
20 30 40
138
RADIO NOISE
75
Fam + D + D
70
Fam + D
65
Fa (dB)
60
Fam
Fam + D D
55
Fam
Fam Dl + Dl
Fam Dl
50
45
40
35
.1 .2
.5
Rural
Quiet Rural
Characteristics
Areas where predominant usage is for any type
of business
Areas used predominantly for single or multiple
family dwellings (at least five single-family
units per hectare), no large or busy highways
Areas where dwelling density is no more than
one every two hectares
No definition given
2.9 107
Fa (dB)
2.9 106
2.9 105
2.9 104
20
2.9
2.9 102
0
0.1
103
5 10 20 30 50 70 80 90 95 99
Percent of time Fa is exceeded
99.9
Temperature (K)
Business
Residential
Rural
Quiet rural
Galactic
40
10
20
30 40 50 60 70 80
Percent exceeded
90
95
98 99 99.5 99.9
99.8
L
Business
L
Residential
L
Rural
0.25
0.50
1.00
2.50
5.00
10.00
20.00
48.00
102.00
250.00
6.1
8.2
2.3
9.1
6.1
4.2
4.9
7.1
8.8
3.8
3.5
4.3
2.5
8.1
5.5
2.9
4.7
4.0
2.7
2.9
3.9
4.4
7.1
8.0
7.7
4.0
4.5
3.2
3.8
2.3
RADIO NOISE
40
Noise power (dB above kT0b)
areas, have decreased dramatically over the years. After reviewing more recent measurements and trend analyses,
Spaulding concluded (23) that in the business environment
at 100 MHz in the 1970s time-frame, Fam was on the order
of 20 dB but now is probably approximately 20 dB less. This
conclusion, however, is not based on a comprehensive set of
noise measurements as would be necessary to update the previous survey described in Ref. 23.
While the improvements in automobile ignition systems
have likely affected the noise levels in business and residential environments, emissions from gap discharge and corona
in power transmission and distribution lines have probably
not decreased with time. Figure 10 (22) shows Fam under, and
one-quarter mile from, a 115 kV line in rural Wyoming. It is
interesting to note that the noise measured one-quarter mile
from the power line is about the same as that predicted for a
rural environment. A possible conclusion is that if power and
distribution lines are the primary noise source in rural environments, rural man-made noise is not likely to have decreased. However, one would not expect noise in an urban
environment to be less (than rural), as would be the case with
the estimated 20 dB reduction in Fam.
Another factor that could significantly affect the level and
character of man-made radio noise is the proliferation of electronic devices (e.g., computers, electronic switching devices,
microwave ovens, etc.) that are unintentional RF emitters.
Such devices have become ubiquitous in business, residential,
and rural environments and could affect both the magnitude
of the noise power as well as its frequency dependence.
The man-made noise data presented in the previous sections are applicable to North America; the validity of extension to other parts of the world cannot be determined precisely. CCIR Report 258 describes very high frequency (VHF)
measurements made in business and residential areas of the
United Kingdom where the noise power was found to be some
10 dB below that shown in Fig. 3 (16). This is attributed to
differences in patterns of utilization of electric and mechanical appliances and regulation of interference. The report also
states that due to such differences, the noise statistics should
be used with caution. It should be noted, however, that if an
overall 10 dB reduction in urban noise can be justified, the
30
20
10
12
15
Hour (MST)
18
21
24
115 kV lineWyoming
wooden towers
100
80
dB above kT0b
30
Fam (dB)
Figure 11. Median, mean, and peak noise power near an office
park (24).
120
60
40
139
20
10
Rural areas
20
0
0.1 0.2
0.5
5 10 20
50 100 200
Frequency (MHz)
500 1000
0
0.1
5 10 20 30 50 70 80 90 95
Percent exceeding ordinate
99
99.9
140
RADIONUCLIDE IMAGING
the locations. Location variations however are quite large, exceeding 12 dB in some cases. More importantly, these measurements show that there are business environments (downtown urban areas) where Fam is still nearly 20 dB.
In summary, the 137 MHz measurements demonstrate
that important changes have occurred in both the level and
character of man-made noise since the comprehensive noise
survey described by Spaulding and Disney (22). While these
measurements can only be considered as a spot check, they
do show that standard methods used to predict man-made radio noise are probably outdated. It is concluded that additional comprehensive man-made noise measurements at RF
frequencies into the ultrahigh frequency (UHF) band will be
necessary to provide radio system designers and engineers
with the required tools to effectively design modern radio
systems.
19. National Bureau of Standards, Quarterly radio noise data, Technical Note 18 (132), US Dept. of Commerce, Washington, DC,
19591966.
20. D. L. Lucas and J. D. Harper, A numerical representation of
CCIR Report 332: High frequency (3-30 Mc/s) atmosphere-radio
data, NBS Tech. Note 318, US Dept. of Commerce, Washington,
DC, 1965.
21. A. D. Spaulding and F. G. Stewart, An Updated Noise Model for
Use in IONCAP, US Dept. of Commerce, Boulder, CO: NTIA Rep
ort 87-212, May 1997.
22. A. D. Spaulding and R. T. Disney, Man-Made Radio Noise Part
1: Estimates for Business, Residential, and Rural Areas, US Dept.
of Commerce, Boulder, CO: OT Rep., 74-38, June 1974.
23. A. D. Spaulding, The roadway natural and man-made noise environment, IVHS J., 2: 175211, 1995.
24. R. Dalke et al., Measurement and analysis of man-made noise in
VHF and UHF bands, 1977 Wireless Commun. Conf., Boulder,
CO, Aug. 1113, 1997, pp. 229233.
BIBLIOGRAPHY
ROGER DALKE
1. W. R. Bennett, Electrical Noise, New York: McGraw-Hill, 1960,
Chap. 5.
2. J. B. Johnson, Thermal agitation of electricity in conductors,
Phys. Rev., 32: 97109, 1928.
3. H. Nyquist, Thermal agitation of electric charge in conductors,
Phys. Rev., 32: 110113, 1928.
4. C. Kittel and H. Kroemer, Thermal Physics, San Francisco, CA:
Freeman, 1980, p. 91.
5. R. Pettai, Noise in Receiving Systems, New York: Wiley, 1984,
Chap. 7.
6. CCIR, Operating Noise-Threshold of a Radio Receiving System,
CCIR Rep. 413, Geneva, Switzerland: International Telecommunications Union, 1966.
7. IRE standards on electron tubes: Definition of terms, 1957, Proc.
IRE, 45: 9831010, 1975.
8. H. T. Friis, Noise figures of radio receivers, Proc. IRE, 32: 419
422, 1944.
9. R. Pettai, Noise in Receiving Systems, New York: Wiley, 1984,
Chap. 11.
10. J. D. Kraus, Radio Astronomy, Powell, OH: Cygnus-Quasar, 1986,
Chap. 3.
11. CCIR, Worldwide Minimum External Noise Levels, 0.1 Hz to 100
GHz, Rep. 670, Geneva, Switzerland: International Telecommunications Union, 1990.
12. CCIR, Characteristics and Applications of Atmospheric Radio
Noise Data, Rep. 332-3, Geneva, Switzerland: International Telecommunications Union, 1986.
13. CCIR, Man-Made Radio Noise, Rep. 258-5, Geneva, Switzerland:
International Telecommunications Union, 1990.
14. CCIR, Radio Emission from Natural Sources in the Frequency
Range above About 50 MHz, Rep. 720-2, Geneva, Switzerland:
International Telecommunications Union, 1986.
15. A. D. Spaulding, The Natural and Man-Made Noise Environment
in Personnel Communications Services Bands, US Dept. of Commerce, Boulder, CO, NTIA Rep. 96-330 (revised), May 1997.
16. R. Achatz et al., Man-Made Noise in the 136138-MHz VHF Meteorological Satellite Band, US Dept. of Commerce, Boulder, CO:
NTIA Rep. 98-355, Sept. 1998.
17. H. Volland, CRC Handbook of Atmospherics, Boca Raton, FL:
CRC Press, 1982, p. 294.
18. T. H. Shumpert, M. A. Honnel, and G. K. Lott, Jr., Measured
spectral amplitude of lightning sferics in the HF, VHF, and UHF
bands, IEEE Trans. Electromagn. Compat., EMC-24: 368369,
1982.
US Department of Commerce
Institute for Telecommunication
Sciences
IONOSPHERE CHARACTERISTICS
The ionosphere poses an interesting challenge for many radio systems that make use of signal transmission
through all or some portion of that medium. Being a magnetoionic medium imbedded in a background neutral
atmosphere, it exhibits very interesting refractive properties, including anisotropy, dispersion, and dissipation.
The laminar ionosphere introduces an array of effects, which are related to the ionospheric component of radio
refractivity. These include ray path bending, phase path increase, group path delay, absorption, Doppler shift,
pulse dispersion, Faraday rotation, and magnetoionic path splitting. Inhomogeneities in the ionosphere give
rise to temporal and spatial variations in the effects just cited. An understanding of the ionospheric personality
provides information about a wide range of solar-terrestrial interactions, and it has significant space-weather
implications. Space weather is a new discipline that includes a wide range of exoatmospheric phenomena of
major importance to space systems and their operational effectiveness.
The main features of the ionosphere are well known, although details are subjects of continuing research.
There are many excellent sources of information about the ionosphere, from both a theoretical and an experimental perspective. The books by Davies (1,2,3), Ratcliffe (4), and Giraud and Petit (5) should be consulted.
Theoretical and plasma-physics aspects of the ionosphere have been discussed in a book by Kelley (6). A readable account of the basic physics of the ionosphere has been given by Rishbeth (7). Other useful references,
which place the ionosphere within a larger context of the geospace weather system, include the Air Force
Handbook of Geophysics and the Space Environment (8), and an Introduction to the Space Environment by
Tascione (9). Various techniques for probing the ionosphere have been described in a monograph by Hunsucker
(10). From a practical perspective, Goodman (11), Johnson et al. (12), and McNamara (13) have published expositions on the ionosphere in connection with radio system applications. There are also proceedings of topical
conferences and workshops. The Ionospheric Effects Symposia (14) have chronicled ionospheric research activities and applications since 1975; and the Commission of the European Communities has published reports
dealing with ionospheric prediction and modeling (15,16).
The purpose of this article is to provide a general understanding of the ionosphere. The emphasis is on
those ionospheric processes and phenomena that are encountered by users of radio propagation systems. More
complete descriptions of the underlying physical processes may be found in various references cited in the text.
A final section on the current status of ionospheric research is provided as an aid to specialists and graduate
students.
IONOSPHERE CHARACTERISTICS
Fig. 1. Depiction of the ionospheric layers and the diurnal variation. [By permission of J. M. Goodman and Kluwer
Academic Publishers, Norwell, MA (11).]
content of the ionosphere (TEC), even though both ionospheric and protonospheric electrons contribute to the
integral. For the purpose of this article, we shall use the more restricted definition for the ionosphere, generally
placing the upper limit at approximately 1000 km. While there are equal numbers of free electrons and positive ions within the ionosphere, it is the electron number density that characterizes the array of interesting
phenomena associated with the region. The ionosphere is imbedded in the earths magnetic field, and this
situation influences the distribution of the ionized constituents. A clear indication of this may be seen in the
worldwide distribution of electron density in the upper ionosphere, which tends to be described by geomagnetic
rather than geographic coordinates. Moreover, being a magnetoionic medium, the ionosphere has a profound
effect upon radiowaves that interact with the medium.
The ionospheric electron density distribution is logically evaluated first in terms of its height profile,
followed by its geographical and temporal variabilities. Though there is abundant evidence suggesting a
rather complex electron density profile comprised of several peaks and valleys, the basis for understanding
fundamental properties of the ionosphere comes from a simple picture of an ionized medium dominated by a
single region, or layer, having a distinct maximum in electron density. This is not without justification, since the
highest and thickest component region, the so-called F layer, typically exhibits the greatest electron density.
Moreover, in many radiowave applications, it is the F layer that exhibits the dominant interaction. Figure 1
IONOSPHERE CHARACTERISTICS
Fig. 2. Various atmospheric and ionospheric layers, the depth of penetration of solar radiation, and the thermospheric
temperature profile. [From a National Research Council report (62).]
depicts the various regions or layers of the ionosphere in terms of the electron number density. It has been
observed that the height profile varies diurnally, seasonally, and as a function of solar activity.
Formation of the Ionosphere. The sun exerts a number of influences on the upper atmosphere, but
the interactions of most importance for our discussion are photodissociation and photoionization. Figure 2
depicts the neutral atmosphere, its various regions, and the depth of penetration of the various components of
solar flux.
In the lower atmosphere, species such as N2 and O2 dominate the constituent population, though other
species such as water vapor, carbon dioxide, nitric oxide, and trace element gases are influential in specific
contexts. In the upper atmosphere, however, molecular forms are dissociated by incoming solar flux into
separate atomic components. Formally the lowest portion of the ionosphere is the so-called D layer at an
altitude of 60 km 20 km, but the free-electron and ion population rises dramatically at an altitude of
100 km, which is the median altitude of the E layer. Two things occur at this altitude. First, oxygen becomes
dissociated as a result of solar UV radiation. Secondly, the mixing of the atmosphere, so efficient below 100 km,
ceases rather dramatically, and the region where this occurs is called the turbopause.
The process of dissociation is so efficient that we treat the distribution of neutral species in a vast segment
of the upper atmosphere (i.e., above 200 km) as that of a monatomic gas. In the lower atmosphere (i.e., below
IONOSPHERE CHARACTERISTICS
Fig. 3. (a) Profiles of ion concentrations, as a function of height, for daytime conditions. (b) Electron density distributions
for day/night and solar maximum/minimum conditions. [From Jursa (8).]
roughly 200 km), the gas is largely polyatomic, although the transition between the two regimes is rather
gradual between 100 km and 200 km. This has implications for the lifetime of ionelectron pairs created
through photoionization. Also, in the altitude regime above about 200 km and well above the turbopause,
collisions become a rarity, so that mixing of the various species becomes unimportant in comparison with
diffusive forces. As a consequence, diffusive separation occurs, with constituents of the neutral gas seeking
their own height distributions dictated by their atomic masses, the gas temperature, and the acceleration of
gravity. Figure 3(a) shows height profiles of ionic species in the upper atmosphere, and Fig. 3(b) shows typical
IONOSPHERE CHARACTERISTICS
distributions of midlatitude electron density for daytime and nighttime under solar maximum and minimum
conditions.
It may be seen that ionized monatomic oxygen is the majority ion between roughly 180 km and 800 km,
and is wholly dominant between about 200 km and 500 km. Atomic hydrogen ions become important above
500 km, and the region from about 800 km to 2000 km is called the protonosphere. It should also be noted that
above 500 km (i.e., the base of the exosphere), the neutral atmosphere is virtually collisionless and particles
tend to move about freely. On the other hand, electrons and ions in the exosphere are still influenced by the
earths magnetic field and electrodynamic forces.
The electron density distributions in the ionosphere and protonosphere are variable. Because of this,
the boundary between the ionosphere and the protonosphere is not sharply defined, being dependent upon
a number of factors including time of day, season, and solar activity. The protonosphere is often referred to
as the plasmasphere, especially by magnetospheric scientists and those engaged in transionospheric TEC
measurements.
Ionospheric Layering. Table 1 provides information about the various ionospheric layers, the altitude
ranges of each, the principal ionic constituents, and the means of formation. A comment is appropriate here
on the nature of ionospheric layering, with some emphasis on the historical distinctions made between the
words layer and region as they pertain to the ionosphere. Often the terms are used interchangeably, and while
neither is generally preferred, region is the more accurate description. This is because it does not convey
the incorrect impression that sharp discontinuities in electron density exist at well-defined upper and lower
boundaries. This is especially the case for the F region, and to a lesser extent for the D and E regions. From a
historical perspective, the concept of layering derives from the appearance of the ionospheric regions on verticalincidence ionospheric soundings, called ionograms (see the subsection Sounder Measurement Method below).
Furthermore, the alphabetic designation of the ionospheric regions was also based upon the early sounding
studies. On the other hand, there are certain situations for which the restrictive term layer is acceptable. For
example, the normal E region may occasionally be characterized by an electron density profile displaying a
degree of boundary sharpness. Aside from this, the most significant localized concentration of free electrons
in the ionosphere, called sporadic E (or Es), exists as an isolated layer within the boundaries of the normal E
region (see the section Sporadic E below). It is termed sporadic because it exhibits a generally unpredictable
temporal and geographical distribution, and because of its limited geographical extent, it is sometimes referred
to as a sporadic E patch.
As indicated above, the ionosphere is often described in terms of its component regions or layers. These
were the so-called D, E, and F regions. These designations are largely based upon data obtained from crude
sounder (i.e., ionogram) measurements undertaken in the 1920s and 1930s. These early measurements often
exhibited evidence for an additional layer between regions E and F in the daytime ionosphere. This led to the
notion that the F region is actually composed of two distinct regions (F1 and F2 ) having different properties.
The lowest region of the ionosphere, the D region, is important in the characterization of absorption losses
for short-wave systems, and also as a reflecting layer for long-wave communication and navigation systems.
There is also evidence for a bifurcation in the D region, with the upper portion (above 60 km) being produced
by solar flux, and with the lower portion (below 60 km) being produced by galactic cosmic rays.
Ground-based vertical-incidence sounder measurements have provided the bulk of our current information about ionospheric structure (see the subsection Sounder Measurement Method below). Through application of ionogram inversion technology to allow for the radio-wave interaction effects, individual sounder stations
provide information about the vertical distribution of ionization to the altitude of the F2 maximum (i.e., 300 km
to 400 km). In addition, the worldwide distribution of these systems has allowed a good geographical picture
to be developed using sophisticated mapping algorithms. These measurements are somewhat limited in the
characterization of certain features such as the so-called EF valley, and they cannot evaluate ionization above
the F2 maximum. There is also a paucity of data over oceanic regions. Satellite measurements (viz., topside
sounders and in situ probes) have been invaluable in the characterization of the F-region ionization density
IONOSPHERE CHARACTERISTICS
over oceanic regions. Rocket probes and incoherent backscatter radar measurements, which provide a clearer
representation of the true electron density profile, typically reveal a relatively featureless profile exhibiting a
single F-region maximum with several underlying ledges or profile derivative discontinuities. Nevertheless,
a valley of ionization may often be observed between the E and F regions. Ionization above the F2 maximum
may be deduced from satellite probes and Thomson-scattering radars, but a large amount of information has
been derived from total electron content measurements using Faraday rotation or group path measurements of
signals from geostationary satellites or Global Positioning System (GPS) satellites. Hunsucker (10) describes
various ionospheric measurement techniques.
Simple layering occurs as the result of two factors. First, the atmospheric neutral density decreases exponentially with altitude, while the solar ionizing flux density increases with height above sea level. This leads
to the formation of single region for which the ionization rate is maximized, and ultimately results in a layer
having the so-called Chapman shape. This shape is based upon a simple theory advanced by Sidney Chapman
in 1931 (Ref. 17; see Fig. 4). We observe nonetheless a degree of structure in the ionosphere, which suggests
more than one layer. One cause for multilayer formation is the existence of a multicomponent atmosphere,
each component of which possesses a separate height distribution at ionospheric altitudes. But there are other
factors. Solar radiation is not monochromatic as suggested in the simple Chapman theory, and it has an energy density that is not evenly distributed in the wavelength domain. Furthermore, its penetration depth and
ionization capability depend upon wavelength and atmospheric constitution. All of this results is a photoionization rate, and an associated electron density profile, that are structured functions of altitude. It has been
shown that the Chapman model is valid for the D, E, and F1 regions but is not generally valid for the F2 region.
Chapman Layer Theory. One of the basic tenets of Chapman theory is that solar radiation will
penetrate to an altitude for which the total number of atoms or molecules, P (populating a column of unit
cross sectional area directed toward the sun) is equal to the reciprocal of the absorption (or interaction) cross
IONOSPHERE CHARACTERISTICS
Fig. 4. An idealized representation of ionization production in the atmosphere as the solar radiation encounters a neutral
gas with exponentially increasing density. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
section a that is P = 1/a. The peak in ionization will be produced in the neighborhood of that altitude, and the
concept is valid for oblique solar illumination as well as for the case in which the sun is directly overhead. It
is convenient to look at the production rate in terms of its deviation from the peak (overhead) value at height
h0 . For this it is useful to define a reduced height z, corresponding to the normalized departure of an arbitrary
value of ionospheric height h from h0 .
where h0 is the peak height for vertically incident radiation from the sun, and H is the neutral scale height
given by the following expression:
where k is Boltzmanns constant, T is the absolute gas temperature, m is the atomic or molecular mass, and g
is the acceleration of gravity. Within the thermosphere (with h 100 km), the gas temperature is monotonically
increasing, reaching an asymptotic level near the base of the exosphere. The temperature rises from 180
K at the mesopause (and incidentally near the turbopause) to levels approaching a diurnal range of 600 K to
1100 K at solar minimum and 800 K to 1400 K at solar maximum. The heat sources include solar radiation,
the dissipation of atmospheric gravity waves, and particle precipitation. The asymptotic levels of T are due to
limits on the thermal conductivity of the gas. The scale height H is a convenient parameter, since it may be
used as a measure of layer thickness for an equivalent fixed-density slab. More importantly, it has a physical
meaning. If the atmosphere is in diffusive equilibrium governed by the force of gravity and the gas pressure
IONOSPHERE CHARACTERISTICS
gradient, and N is the atomic or molecular gas density (as appropriate), we have
At altitudes well above the peak in q, the rate of electron production drops off in an exponential fashion
imitating the exponential decrease in gas pressure with height. In order to relate Chapman production curves
to actual electron density distributions, we must examine loss processes and certain dynamic factors.
where N e is the electron density, L(N e ) is the loss rate, which is dependent upon the electron density, div stands
for the vector divergence operator, and V is the electron drift velocity.
The divergence of the vector in Eq. 6 is the transport term, sometimes conveniently called the movement
term. The continuity equation says that the time derivative of the electron density within a unit volume is equal
to the number of electrons that are generated within the volume (through photoionization processes) minus the
number that are lost (through chemical recombination or attachment processes), and finally adjusted for those
electrons that exit or enter the volume (as expressed by the transport term). To first order, the only derivatives
of importance in the divergence term are in the vertical direction, since horizontal N e gradients are generally
smaller than vertical ones. In addition, there is a tendency for horizontal velocities to be small in comparison
with vertical drift velocities. Consequently, we may replace div (N e V) by (d/dh) (N e V h ), where V h is the scalar
IONOSPHERE CHARACTERISTICS
Fig. 5. Curve illustrating the rate of electron production as a function of reduced height (h h0 ) and for selected values
of the solar zenith angle. [From Davies (1).]
10
IONOSPHERE CHARACTERISTICS
Now let us look at some special cases. If V h = 0 (no movement), then the time variation in electron
concentration is controlled by a competition between production q and loss L. At nighttime, we may take q =
0, and this results in
In principle, there are two mechanisms to explain electron loss: attachment of electrons to neutral atoms
(in the upper ionosphere), and recombination of electrons with positive ions (in the lower ionosphere). The
attachment process is proportional to N e alone, while recombination depends upon N e with N i , where N i is the
number of ions. Attachment involves radiative processes and has an extremely low cross section (probability
of occurrence). We may ignore it in many practical situations and take recombination as the major source
for electron loss. Since N e = N i , the recombination process obeys the equation L = N e 2 , where is the
recombination coefficient. Recombination is very rapid in the D and E regions, the process being accomplished
in a time on the order of seconds to minutes. Attachment, the electron loss process for the upper ionosphere, has
a time constant on the order of hours. This is the primary reason that the ionosphere does not entirely disappear
overnight. Another reason is that there exists a second source of electrons associated with the plasmasphere.
This reservoir of ionization is built up during the daytime through vertical drift, but bleeds into the ionosphere
during nocturnal hours.
In the vicinity of local noon, dN e /dt = 0 and we may analyze the quasiequilibrium conditions suggested
by Eq. (7) when the left-hand side of the equation equals 0. The two main types of equilibrium processes are
given in Table 2.
The equilibrium processes identified in Table 2 are the dominant possibilities during daytime when
photoionization is significant. During nocturnal hours, equilibrium is seldom achieved at F-region heights,
although it is approached in the period before sunrise.
While the continuity equation appears quite simple, the generic terms (i.e., production, loss, and transport)
represent a host of complex photochemical and electrodynamic processes, which exhibit global variations and
are influenced by nonstationary boundary conditions within the atmosphere and the overlying magnetosphere.
Notwithstanding these complications, the equation provides a remarkably clear view of the basic processes
that account for ionospheric behavior. In fact, the relative contributions of terms in the continuity equation
will account for the majority of the anomalous ionospheric properties; that is, those ionospheric variations that
depart from a Chapman-like characteristic. This is especially true for the F2 layer, within which the movement
term attains paramount status. In the E and F1 regions, where the movement term is small compared with
production and loss (through recombination), photochemical equilibrium exists in the neighborhood of midday.
All of this has had a significant bearing on the development of ionospheric models and prediction methods.
IONOSPHERE CHARACTERISTICS
11
Indeed, as it relates to the F region of the ionosphere, it may be said that the existence of a nonvanishing
divergence term in the continuity equation has been the primary impetus for the development of statistical
modeling approaches. Nevertheless, efforts to account for all terms in the continuity equation through physical
modeling are ongoing.
The underlying assumptions used by Chapman in his theory of layer production are in substantial
disagreement with observation. The Chapman layer was based upon an isothermal atmosphere, and it is well
known that the atmosphere has a scale height, kT/mg, which varies with height. Moreover, the basic theory
assumes a monochromatic source for photoionization and a single constituent gas. Corrections and extensions
to the early Chapman theory have led to better agreement with observation, and to this day the Chapman
layer provides a fundamental baseline for ionospheric profile modeling.
12
IONOSPHERE CHARACTERISTICS
Fig. 6. Typical vertical-incidence ionosonde recording (i.e., ionogram) and the corresponding plasma frequency profile
fp(h). The electron density profile is related to the plasma frequency profile by the Eq. 9 in the text. [This ionogram was
derived from U.S. Government web site (http://www.ngdc.noaa.gov/stp/), and the instrument was developed by University
of Massachusetts-Lowell for the U.S. Air Force.]
expression is used:
where f O is the ordinary-ray critical frequency (MHz) and N max is the maximum electron density of the given
layer (e/m3 ). Equation (10) is equivalent to Eq. (9).
From a historical perspective, it is interesting to note that the concept of radar detection of aircraft
derived from the early work of ionospheric specialists who were already using ionospheric sounders as a means
to detect ionospheric layers.
The D Region. The D region is responsible for most of the absorption encountered by HF signals, which
exploit the sky-wave mode. In most instances, D-region absorption is a primary factor in the determination
of the lowest frequency, which is useful for communication over a fixed sky-wave circuit. In addition, the D
region supports long-wave propagation at very low frequency (VLF) and low frequency (LF), and the medium
is exploited in certain legacy navigation systems and strategic low-rate communication systems. The sounder
method as described in the previous section is not useful for measurement of the D region, since the electron densities are relatively low. Details of D-region electron concentration are sketchy in comparison with
IONOSPHERE CHARACTERISTICS
13
information available about the E and F regions, principally because of the difficulty in making diagnostic
measurements. Moreover, analysis is hampered because many photochemical processes with poorly defined
reaction rates take place in the D region. Over 100 reactions have been compiled.
Table 1 shows that the D region lies between 70 km and 90 km. In fact, the upper and lower levels are
not precisely defined. It is evident that more than one source of ionization gives rise to the D-region electron
density distribution. Sources include solar radiation at the upper levels and galactic cosmic rays at lower
levels. In addition, relatively rare polar-cap absorption (PCA) events are characterized by highly energetic
solar protons that provide an additional source for ionization of the lower D region within the polar cap. Some
investigators place the lower boundary of the D region at 50 km to allow for the contribution of galactic cosmic
rays in the neighborhood of 50 km to 70 km. This altitude regime, termed the C region, is not produced by
solar radiation. It exhibits different characteristics from the region between 70 km and 90 km. Specifically, a
minimum in electron concentration is observed during solar-maximum conditions for the lower portion (viz.,
region C), while the reverse is true in the upper portion (viz., region D). This can be explained if we assume
that the galactic cosmic rays are partly diverted from the earth by an increase in the interplanetary magnetic
field (IMF), which occurs during solar maximum conditions.
The E Region. In an () Chapman layer for which photochemical equilibrium has been established,
the following equation represents the electron density distribution as a function of reduced height z:
where is the recombination coefficient, is the solar zenith angle, and q0 is the maximum production rate
in the layer. Recall that is the recombination coefficient (see the section The Continuity Equation and
Equilibrium Processes above). The quantity q0 / is dependent upon the sunspot number and is specific to the
region involved, in this case the E region. The maximum rate of electron production q0 occurs only for the
overhead sun. However, it may be shown that actual maxima for other zenith angles are simply related by this
expression:
It may be shown that the ordinary-ray critical frequency for the E region, which is directly related to the
E-region maximum electron density through Eq. (10), may be found from Eq. (11), and is given by
where is a constant of proportionality, which is dependent upon the sunspot number. The exponent n tends to
a value 0.25 for long-term seasonal behavior, and in compliance with Chapman theory, but some workers have
found that a value for n 0.3 better represents the diurnal dependence. The constant of proportionality ranges
between about 3 MHz and 4 MHz, bearing in mind that Eq. (13) represents a climatological median value.
The solar-activity dependence of the ratio of peak production to the effective loss (recombination) coefficient has been studied by a number of workers, and the results enable values of foE to be deduced. There have
also been direct measurements of foE using vertical incidence sounders. While there is some variability to be
considered, it is possible to develop a relationship connecting the median value of foE, the solar zenith angle,
and the 12-month running-mean sunspot number. A generally accepted candidate for the daytime E-region
critical frequency is
where R12 is the running 12-month sunspot number, which may range between roughly 10 and 150.
Equation (14) provides excellent agreement with observation during the daytime, but alternative expressions are found to be more appropriate during the nighttime hours (3). Moreover, it has been found that
14
IONOSPHERE CHARACTERISTICS
Fig. 7. (a) Depiction of the local-time (LST) and latitude dependence of foE for solar-maximum conditions in summer
[from Davies (1)](b) Contours of foE at Fort Belvoir, Virginia, in 1958 (solar maximum), showing seasonal variations. The
contours are in megahertz.
Eq. (14) is inaccurate in the very high latitudes, where other means of electron production become important,
invalidating the Chapman hypothesis. Internationally adopted relations for monthly median foE are due to
Muggleton (18); and an alternative relation, specific to the European region, has been published (15).
Figure 7(a) contains an E-region critical-frequency map for summer solstice conditions in 1958, a period
of high solar activity (i.e., R12 large). The contours are representative of median conditions as a function of
IONOSPHERE CHARACTERISTICS
15
geographic latitude and local time. It is seen that the E-region critical frequencies (and consequently the
electron densities) are vanishingly small in regions devoid of solar illumination. This summer solstice behavior
is consistent with Eq. (14), and other seasons have been shown to behave in conformance with (cos )0.25 as
well.
Figure 7(b) shows the monthly variation of foE for one station (Ft. Belvoir, Virginia) for the year 1958.
The solar control is obvious in the median data plotted.
The F1 Region. The F1 region is not unlike the E region in the sense that it obeys many of the
predictions of Chapman theory. We look for a relation for the ordinary-ray critical frequency that is formally
similar to Eq. (13). A relation patterned after Chapman principles may be expressed as
Like the E region, the F1 region exhibits more complicated behavior than that expressed by such a simple
formula. Specifically, it has been found that the geomagnetic latitude tends to exhibit some control over the
F1 -region electron densities. The function f s in Eq. (15) depends upon sunspot number and magnetic latitude.
It is also observed that the F1 region disappears (i.e., merges with the F2 region) at values of the solar zenith
angle exceeding a certain maximum that itself depends upon both the sunspot number and the geomagnetic
latitude. The Radio Sector of the International Telecommunications Union (ITU-R, previously the CCIR) has
developed a method for computing foF1 taking all these factors into account (18). The internationally adopted
monthly median foF1 formulation is based on the work of Ducharme et al. (19). The relation due to Davies (3)
gives a convenient but approximate expression for the F1 -layer critical frequency:
Figure 8 shows the solar-zenith-angle control of foF1 under sunspot maximum and minimum conditions.
The height of the F1 ledge, hF1, is taken to be between 180 km and 210 km. From Chapman theory we
anticipate that hF1 will be lower in summer than in winter and will be higher at midlatitudes than at low
latitudes. Unfortunately, the reverse is true. Explanations for this behavior may be found in a detailed study of
scale-height gradients, a nonvanishing movement term (as expressed in the continuity equation), or gradients
in upper atmospheric chemistry.
The F2 Region. The F2 region is the most prominent layer in the ionosphere, and this significance
arises as a result of its height (it is the highest of all the component layers) and of course its dominant electron
density. It is also characterized by large ensembles of irregularity scales {L} and temporal variations {T}.
The F2 region is a vast zone, which eludes prediction on the microscale ( L < 1 km) and mesoscale (1 km <
L < 1000 km) levels, and even provides challenges to forecasters for global and macroscale (L > 1000 km)
variations. This is largely because of the elusive transport term in the continuity equation. There are also a
host of so-called anomalous variations to consider, and these are the subjects of a succeeding section.
As in the E and F1 regions, we may conveniently specify the behavior of the F2 region in terms of equivalent
plasma frequency rather than the electron density. For the peak of ionization we have
16
IONOSPHERE CHARACTERISTICS
Fig. 8. Depiction of the local-time and latitudinal variation of foF1 for two different solar activity conditions: (a) solar
minimum, June 1954; (b) solar maximum, June 1958. The contours are in megahertz. [From Jursa (8).]
The CCIR published its CCIR Atlas of Ionospheric Characteristics, which includes global maps of F2 -layer
properties for sunspot numbers of 0 and 100, for every month, and for every even hour of Universal Time (20).
Figure 9 is an illustration of the global distribution of foF2 for a sunspot number of 100. Such maps are derived
from coefficients based upon data obtained from a number of ionosonde stations for the years 19541958 as
well as for the year 1964. This set of coefficients is sometimes identified by an ITS prefix, but is known more
IONOSPHERE CHARACTERISTICS
17
Fig. 9. Map of foF2 showing the worldwide distribution under the following conditions: 15 November, Sunspot Number =
135, Time = 0000 UTC. The countours of foF2 are developed using the URSI set of ionospheric coefficients. Curves similar
to this are found in the Atlas of Global Ionospheric Coefficients (20). [By permission, Radio Propagation Services, November
2000.]
generally as the CCIR coefficients. Because of the paucity of data over oceanic areas, a method for improving
the basic set of coefficients by adding theoretically derived data points was developed. As a result, a new set of
coefficients has been sanctioned by International Union of Radio Science (URSI), and this is termed the URSI
coefficient set. Many communication prediction codes, which require ionospheric submodels, allow selection of
either set of ionospheric coefficients.
Anomalous Features of the Ionospheric F Region. The F2 layer of the ionosphere is probably the
most important region for many radio-wave systems. Unfortunately, the F2 layer exhibits the greatest degree of
unpredictable variability because of the transport term in the continuity equation. As indicated previously, this
term represents the influences of ionospheric winds, diffusion, and dynamical forces. The Chapman description
for ionospheric behavior depends critically upon the unimportance of the transport function. Consequently,
many of the attractive, and intuitive, features of the Chapman model are not observed in the F2 region.
The differences between actual observations and predictions derived on the basis of a hypothetical Chapman
description have been termed anomalies. In many instances, this non-Chapman-like behavior is not anomalous
at all, but rather typical.
The following list represents the major forms of anomalous behavior in the F2 layer: diurnal, Appleton,
December, winter, and the F-region trough. A few comments are provided for each major form.
The Diurnal Anomaly. The diurnal anomaly refers to the situation in which the maximum value of
ionization in the F2 layer occurs at a time other than at local noon as predicted by Chapman theory. On a
statistical basis, the actual maximum occurs typically in the temporal neighborhood of 1300 to 1500 LMT.
Furthermore, there is a semidiurnal component that produces secondary maxima at approximately 1000 to
18
IONOSPHERE CHARACTERISTICS
1100 LMT and 2200 to 2300 LMT. Two daytime maxima are sometimes observed (one near 1000 and the other
near 1400), and these may cause the appearance of a minimum at local noon. This feature, when observed, is
called the midday biteout.
Appleton Anomaly. This feature is symmetric about the geomagnetic equator and goes by a number of
other names, including the geographic anomaly, the geomagnetic anomaly, and the equatorial anomaly. The
Appleton anomaly is associated with the significant departure in the latitudinal distribution of the maximum
electron concentration within 20 to 30 on either side of the geomagnetic equator. Early in the morning a
single ionization peak is observed over the magnetic equator. However, after a few hours the equatorial F
region is characterized by two distinct crests of ionization that increase in electron density as they migrate
poleward. This phenomenon is described as an equatorial fountain initiated by an E B plasma drift (termed
a Hall drift), where E is the equatorial electrojet electric field and B is the geomagnetic field vector. This drift is
upwards during the day, since the equatorial electric field E is eastward at that time. As the electrojet decays,
the displaced plasma is now subject to downward diffusion when the atmosphere begins to cool. This diffusion
is constrained along paths parallel to B, which map to either side of the geomagnetic equator. The poleward
extent of the anomaly crests is increased if the initial Hall-drift amplitude is large. This anomalous behavior
accounts for the valley in the parameter foF2 (with peaks on either side) seen at the geomagnetic equator in
Figure 9. There are significant day-to-day, seasonal, and solar-controlled variations in the onset, magnitude,
and position of the anomaly. There are also asymmetries in the anomaly crest position and electron density.
Asymmetries in the electron density in the anomaly crests appear to be the result of thermospheric winds that
blow across the equator from the subsolar point. The effect of magnetic activity on the anomaly is to constrain
the electron density and latitudinal separation of the crests. Magnetic activity is monitored worldwide, and the
quasilogarithmic index Kp is used to represent the level of worldwide activity (21). When Kp 5 (on a scale
from 0 to 9), the anomaly disappears.
The December Anomaly. This term refers to the fact that the electron density at the F2 peak over
the entire earth is 20% higher in December than in June, even though the solar-flux change due to earth
eccentricity is only 5% (with the maximum in January).
The Winter (Seasonal) Anomaly. This is the effect in which the noontime peak electron densities are
higher in the winter than in the summer despite the fact that solar zenith angle is smaller in the summer
than it is in the winter. This effect is modulated by the 11-year solar cycle and virtually disappears at solar
minimum.
The F-Region (High-Latitude) Trough. This is representative of a number of anomalous features that are
associated with various circumpolar phenomena, including particle precipitation, the auroral arc formations,
etc. The high-latitude trough is a depression in ionization, occurring mainly in the nighttime sector, and it is
most evident in the upper F region (22). It extends from 2 to 10 equatorward of the auroral oval, an annular
region of enhanced ionization associated with optical aurora (see the section The High-Latitude Ionosphere).
The trough region is associated with a mapping of the plasmapause onto the ionosphere along geomagnetic
field lines (see Fig. 17). The low electron density within the trough results from a lack of replenishment through
candidate processes such as antisunward drift, particle precipitation, or the storage effect of closed field lines.
The latitudinal boundaries of the trough may be sharp, especially the poleward boundary with the auroral
oval. A model of the trough is due to Halcrow and Nisbet (23).
Irregularities in the Ionosphere. In addition to the various anomalous features, irregularities in the
electron density distribution may be observed throughout the ionosphere. The size, intensity, and location of
these irregular formations are dependent upon a number of factors, including geographical area, season, time
of day, and levels of solar and magnetic activity. The traveling ionospheric disturbance (TID; see the subsection
Short-Term Variations and the section Ionospheric Predictions below) belongs to a special class of irregular
formations that are generally associated with significant changes in the electron density (more than a few
percent) over large distances (> 10 km). The remaining irregularities, loosely termed ionospheric inhomogeneities, typically develop as the result of ionospheric instability processes and are not directly associated
IONOSPHERE CHARACTERISTICS
19
with TIDs. On the other hand, TIDs have been shown to be a possible catalyst in the formation of ionospheric
inhomogeneities, especially in the vicinity of the Appleton anomaly. Relatively small-scale ionospheric inhomogeneities are important, since they are responsible for the rapid fading (scintillation) of radio signals from
satellite communication and navigation systems. Such effects may introduce performance degradations or
outages on systems operating at frequencies between 100 MHz and several gigahertz. Models of radiowave
scintillation have been developed, and these are based upon a basic understanding of the global morphology of
ionospheric inhomogeneities.
There are inhomogeneities in all regions of the ionosphere, but the equatorial and high-latitude regions are
the most significant sources. Hunsucker and Greenwald (24) have reviewed irregularities in the high-latitude
ionosphere, and Aarons (25) has examined the equatorial environment.
Equatorial inhomogeneities tend to develop following sunset and may persist throughout the evening, but
with decreased intensity after local midnight. The irregularities are thought to be the result of an instability
brought about by a dramatic change in F-region height at the magnetic equator following sunset. The scale
lengths of the irregularities may range between roughly a meter and several kilometers, and the spectrum of the
irregularities has been observed to exhibit a power-law distribution. There is a tendency for the irregularities to
be field-aligned with an axial ratio of roughly 20 to 1. In addition, the irregularities are organized in distended
patches. Though the situation is variable, the patch sizes range between 100 km and several thousand
kilometers in the upper F region, and average 100 km in the lower F region. The equatorial irregularities
tend to be more intense and widespread at the equinoxes and at solar maximum, but magnetic activity tends
to suppress the growth of the irregularities.
High-latitude irregularities exist within the polar cap and the auroral zone, with the latter being primarily
associated with the bottomside F region. The high-latitude F region is quite variable, and unlike midlatitudes,
it may have an electron density less than the E-region during nocturnal hours. In the wintertime, structured
auroral arcs may migrate within the polar cap, and the electron density enhancements within these formations
may be several orders of magnitude greater than the normal background, especially during elevated solar
activity. During disturbed geomagnetic conditions, structured electron density patches have been observed to
travel across the polar cap in the antisunward direction. These irregularities may have a significant effect on
communication systems. For both the auroral zone and the polar cap, increased geomagnetic activity has a
dramatic influence on the growth of irregular ionospheric formations. Moreover, for large and sustained values
of Kp, it has been observed that the high-latitude irregularity patterns tend to migrate equatorward, replacing
the background midlatitude properties (see the section The High-Latitude Ionosphere).
20
IONOSPHERE CHARACTERISTICS
Fig. 10. Mean diurnal variation of foE, foF1, and foF2 for summer and winter under northern-hemisphere- and solarmaximum conditions. [From Jursa (8).]
nonuniform changes in atmospheric pressure. Figure 12 shows the variation of foF2 as a function of time,
showing the impact of TIDs. Figure 13 shows the effect of a large geomagnetic storm.
IONOSPHERE CHARACTERISTICS
21
Fig. 11. Variations in the hourly values of foF2 as a function of the time of day, for January solar maximum conditions at
a northern-hemisphere midlatitude site. The range of day-to-day variability in foF2 is 10%, suggesting a variation in
NmaxF2 of 5%. [From Davies (1).]
Sporadic E
General Characteristics. Even though the normal E region is Chapman-like in nature, isolated forms
of ionization are often observed in the E-region, having a variety of shapes and sizes. These ionization forms
have been termed sporadic E, because they appear quasirandomly from day to day, and they generally defy
deterministic prediction methods. Sporadic E (Es) ionization has been observed during rocket flights and with
incoherent backscatter radar, and a layer thickness of the order of 2 km has been observed. It generally takes
the form of large-scale structures, having horizontal dimensions of hundreds of kilometers at middle latitudes.
22
IONOSPHERE CHARACTERISTICS
Fig. 12. Variations in the ionosphere thought to be associated with traveling ionospheric disturbances. The foF2 variations
shown here are of the order of 2% and have periods of 20 min. The NmaxF2 variations are 1%. [From Paul (63).]
Fig. 13. Effect of a large geomagnetic storm on N max . [By permission of J. M. Goodman and Kluwer Academic Publishers,
Norwell, MA (11).]
IONOSPHERE CHARACTERISTICS
23
Fig. 14. Variation in R12 , foF2, foE, and 4 MHz absorption at noontime. The seasonal effects are clearly evident, the
foE and D-layer variations being out of phase with the foF2 variations (i.e., seasonal anomaly). [By permission of J. M.
Goodman and Kluwer Academic Publishers, Norwell, MA (11).]
Polar and equatorial forms have different structures and causal mechanisms. Although sporadic E consists
of an excess of ionization (against the normal E-region background), it does not appear to be strongly tied
to solar photoionization processes. Still, midlatitude Es occurs predominantly during summer days. Sporadic
E does exhibit seasonal and diurnal tendencies, which have been examined statistically, and at least three
different types of sporadic-E ionization have been discovered with distinct geographical regimes: low-latitude
24
IONOSPHERE CHARACTERISTICS
Fig. 15. Long-term variation in R12 , foF2, and foE at noontime. Since running 12-month averages were taken, the seasonal
effects observed in Fig. 14 are smoothed out. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
(or equatorial), midlatitude (or temperate), and high-latitude ionization. Figure 16 depicts the probability of
Es occurrence.
Formation of Midlatitude Sporadic E. It has suggested that wind-shears in the upper atmosphere
are responsible for the formation of sporadic E at midlatitudes. We shall review this process briefly.
It should be recalled from the examination of photochemistry in the ionosphere that molecular ions such
as those that exist in the E region introduce rapid electron loss by recombination. At the same time it is
recognized that an enormous number of meteors burn up in the E region. This meteoric debris is largely
comprised of metallic ions, which are monatomic. Their presence has been confirmed by mass spectroscopy
measurements using rockets, and they include iron, sodium, magnesium, etc. Since monatomic ions exhibit a
small cross section for electron capture, the process by which atomic ions become concentrated in well-defined
layers will lead to reduced loss rates for ambient free electrons in the interaction region.
The influx of this foreign mass of metallic ions, when distributed over the whole of the E region, would
be insufficient to overwhelm the omnipresent molecular species (such as NO+ ), which are in a state of photochemical equilibrium, were it not for a mechanism that preferentially concentrates the meteoric debris ions.
Apparently wind shear is this mechanism. The basic wind shear theory was proposed by Whitehead (26), but
it remained for Gossard and Hooke (27) to outline a process for meteoric ion concentration based upon the
interaction of the meteoric debris with atmospheric gravity waves, the latter wave structures being responsible
for the development of TIDs as well. The ultimate process involves a corkscrew propagation of atmospheric
gravity waves and atmospheric tides, which results in a rotation of wind velocity as a function of altitude. This
IONOSPHERE CHARACTERISTICS
25
Fig. 16. Probability of Es occurrence as observed in the period 19511952. It is representative of the global, seasonal, and
diurnal variation of sporadic-E ionization. [From Davies (1).]
effect can cause the wind to change direction over an altitude of only a kilometer or so, so as to trap meteoric
ions at an intermediate point having zero velocity. This buildup in a narrow region is sufficient to generate an
intense sporadic-E patch.
Sporadic E at Non temperate Latitudes. The high-latitude sources are evidently of two types,
depending upon whether the observation is made in the neighborhood of the auroral oval or poleward of it
26
IONOSPHERE CHARACTERISTICS
Fig. 17. Idealized picture of ionospheric plasma frequencies in a northsouth plane through Fairbanks and Anchorage,
Alaska. E, region equatorward of trough; B, equatorward edge of trough: C, plasma frequencies (MHz); D, trough minimum;
E, plasmapause field line; F, poleward edge of trough; G, F-region blobs; H, enhanced D-region absorption; I, E-region
irregularities. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11), after Hunsucker (28).]
(i.e., in the polar cap region). It has been found that auroral Es is basically a nocturnal phenomenon; it is
associated with the optical aurora and is due to auroral electron precipitation. Because of its proximity to the
seat of auroral substorm activity, it is not surprising to find some correlation between auroral Es and some
appropriate magnetic index. Indeed, it has been found that auroral Es is positively correlated with magnetic
activity. On the other hand, polar-cap Es may be relatively weak, and is negatively correlated with substorm
activity.
Turning equatorward, it has been found that equatorial Es is most pronounced during daylight hours, and
evidence points to the formation of ionization irregularities within the equatorial electrojet as the responsible
agent at low latitudes.
IONOSPHERE CHARACTERISTICS
27
Fig. 18. Depiction of various ionospheric features at a given time such that the daynight terminator is passing through
the middle of the United States (i.e., 2300 GMT). [By permission of J. M. Goodman and Kluwer Academic Publishers,
Norwell, MA (11), from Bishop et al. (29).]
The magnetic activity index K p is generally available and is typically used as the parameter of choice
to determine the statistical position of the auroral zone. The concept of the auroral oval was developed by
Feldstein and Starkov (30) on the basis of a set of all-sky camera photographs that were obtained during the
International Geophysical Year. Other models exist, but the Feldstein picture is found in most models that
attempt to include auroral effects in some way. The position of the oval is important, not only as an ionospheric
feature in itself, but because it also represents a boundary between the decidedly different geophysical regimes
that are poleward of it (the polar cap) and equatorward of it (the midlatitudes). Because the position of the
auroral zone varies diurnally as well as with the index K p , there are some sites that may be characterized by
all four regimes at any given time: polar, auroral, trough, and midlatitude. Iceland is such a location.
One of the most fascinating properties of the various circumpolar features is their latitudinal motion as
a function of magnetic activity. The ionospheric plasma is best organized in terms of some form of geomagnetic
coordinates, but the high-latitude plasma patterns are not fixed in that frame of reference either. The equatorward boundary of the region of precipitating electrons has been deduced from DMSP satellite instruments,
and it takes a form due to Gussenhoven et al. (31):
where corrected geomagnetic coordinates are used, L(t) and L0 (t) are specified in degrees, and L0 (t) is the
equatorward boundary of the oval when K p = 0. It is emphasized that L0 (t) and a(t) are functions of time
in the magnetic local time (MLT) system.. Both functions are smoothly varying over the diurnal cycle: L0
ranges between 65 at 0100 MLT and 72 at 1700 MLT, and a(t) varies between 2 at 2400 MLT and
28
IONOSPHERE CHARACTERISTICS
Fig. 19. (a) Descent of the auroral oval as a function of magnetic activity; (b) position of the auroral oval and its thickness
versus the magnetic index K p ; (c) position of the auroral arc formations versus the magnetic activity index Dst . [By
permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11).]
0.8 at 1500 MLT. Therefore the statistical representation of the oval has its greatest equatorward descent
during nocturnal hours. Moreover this equatorward boundary is greatly influenced by magnetic activity. Chubb
and Hicks (32) have found that the daytime aurora descends approximately 1.7 degrees per unit K p , and the
nighttime aurora descends at a rate of 1.3 degrees per unit K p . The auroral oval and thickness are depicted in
Fig. 19. Ultimately the auroral arcs, which reside within the auroral oval, are tied to interplanetary phenomena.
IONOSPHERE CHARACTERISTICS
29
Fig. 20. Hierarchy of solarterrestrial effects. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
Workers have shown that the magnitude of the southward component of the interplanetary magnetic field is a
key factor is the development of so-called geomagnetic substorms, wherein K p exhibits large enhancements.
The US Air Force prepares daily summaries of an index Q in order to provide a basis for various analyses of
the high-latitude ionosphere. The index Q ranges between 0 and 8, with larger values associated with a widening
of the oval region and a general increase in intensity of activity within the oval. Moreover, the equatorward
boundary of the auroral oval moves to lower latitudes as Q increases. Since Q, viewed as a parameter, defines
the shape and location of the auroral zone, it is a convenient index for transmission to communication facilities
and forecasting facilities. Its utility is dependent upon timeliness and accuracy. As originally designed by
Feldstein, Q defines only a statistical relationship between the oval position and magnetic activity, the latter
being parameterized by the planetary index K p . Nevertheless, the Feldstein oval concept has been shown to
have some utility under real-time circumstances. Satellite imagery is used to deduce an effective Q.
Auroral physics is an exceedingly rich and complex subject. Not all phenomena in the high-latitude region
are understood, and insufficient data are available to fully characterize even those factors for which a general
understanding exists.
30
IONOSPHERE CHARACTERISTICS
a significant increase in D-layer ionization with a temporal pattern that mimics the flare itself. This results in
an increase in the product of the electron density and the collision frequency. It is the growth of this product
that accounts for the absorption of HF signals passing through the D region. Flares tend to be more prevalent
during the peak in sunspot activity, and the individual-flare duration distribution ranges from a few seconds
to roughly an hour.
IONOSPHERE CHARACTERISTICS
31
Ionospheric Models
As in many areas of geophysical study, ionospheric modeling may assume a number of forms, ranging from the
purely theoretical to the totally empirical. Approaches may also include a combination of these forms, although
empirical models dominate the field. Recent developments include allowance for adaptivity within the models
to accommodate exploitation in the near-real-time environment for special applications. While physical or
theoretical principles are the inspiration for a number of models, in fact most models in use today are largely
specified on the basis of semiempirical relationships derived from observational data.
Ionospheric models fulfill a variety of needs beyond basic research, with the most prominent application
being the assessment and prediction of radio-system performance. For example, ionospheric models are the
engines that drive HF system performance models such as IONCAP (37). Related models are supported by the
U.S. Department of Commerce, including VOACAP, ICEPAC, and REC533 (38). Other applications include evaluation of transionospheric signal parameters and errors in ranging or geolocation introduced by the electron
content of the ionosphere. A general discussion of the status of ionospheric modeling in the context of HF communication systems has been covered by Goodman (11), and recent information regarding telecommunication
system planning has been published by the Commission of European Communities (16).
Ionospheric profile models are based upon the superposition of various submodels of the ionospheric layers or regions (i.e., D, E, Es, F 1 , and F 2 ).The basic purpose of modeling is to represent the electron density
profile under a variety of conditions [see Fig. 3(b)]. These submodels may represent the respective layers as
thin horizontal sheets (e.g. sporadic E) or quasiparabolic regions in the vicinity of maximum ionization. The
submodels are specified by the maximum electron density of the layer, the layer height, the layer thickness,
and a functional representation of the layer shape. There are a number of models for the height profile, the
main differences being the manner in which the component layers are combined. Figure 21 depicts the International Reference Ionosphere (39) and the ionospheric model in the computer program IONCAP. There
are also geographical, seasonal, and solar epochal variations in the specified ionospheric profiles and the
parameters upon which they are built. An example of the geographical variations in foF2 was shown in
Fig. 9, and the Global Atlas of Ionospheric Coefficients was discussed in the section Description of the Ionospheric Layers. Ionospheric coefficients used to produce maps similar to Fig. 9 are common to virtually all
global models of the ionosphere. Currently there are two sets of ionospheric coefficients that may be specified:
the original CCIR (or ITU-R) set, which is sanctioned by the ITU-R, and the newer URSI set (40).
A simplistic model of the ionosphere consists of a parabolic E layer, a linear increase in electron density in
the F 1 layer a parabolic F 2 layer (41). At nighttime, the E and F 1 layers effectively disappear. A newer ITU-R
recipe, consisting of multiple quasiparabolic layers to provide continuity of the overall profile and its height
derivatives (42), has replaced this so-called BradleyDudeney profile model.
Significant improvements in empirical ionospheric modeling have been promoted by military agencies
around the world, including the US Department of Defense, the UK Ministry of Defence, and others. This is not
surprising in view of the large number of applications of ionospheric specification in radio-wave systems used
by the military. The original ICED model was intended to be a northern-hemisphere ionospheric specification
model to serve the requirements of the US Air Force. It was only a regional model, descriptive of midlatitude
behavior but extending into the auroral zone. It was designed to allow for recovery of some of the dynamic
features embodied in auroral climatology that are smeared out in most mapping procedures. The model, as
described by Tascione et al. (43,44), is driven by an effective sunspot number and an index derived from auroral
oval imagery. The effective sunspot number is not based on solar data at all, but is derived from ionospheric
data extracted from the US Air Force real-time ionosonde network. This sunspot number is similar to an
ionospheric T index developed by Australian workers, and to the pseudoflux concept used by the US Navy
for HF predictions (11). The ICED model has been generalized to incorporate global considerations, while
emphasizing near-real-time applications. Anderson et al. have developed a low-latitude ionospheric profile
model, SLIM (45) and a fully analytic ionospheric model, FAIM (46) in order to eliminate the use of limiting
32
IONOSPHERE CHARACTERISTICS
Fig. 21. (a) International Reference Ionosphere (IRI); (b) IONCAP ionospheric submodel. [By permission of J. M. Goodman
and Kluwer Academic Publishers, Norwell, MA (11).]
simplifications in the driving parameters associated with prediction models. A discussion of SLIM and FAIM
may be found in a paper by Bilitza (47). Other developments supporting Air Force requirements include PIM
and PRISM. PIM (Parameterized Ionospheric Model) is a global model of theoretical and empirical climatology,
which specifies the ionospheric electron and ion densities from 90 km to 25,000 km. PRISM (Parameterized
Real-Time Ionospheric Specification Model) uses ground-based and space-based data available in real time
IONOSPHERE CHARACTERISTICS
33
to modify PIM, thereby providing a near-real-time ionospheric specification. Another model, RIBG, developed
by Reilly (48), combines ICED and several other models with a general ray-tracing utility. Current versions
of these models and validation of PRISM are discussed by Doherty et al. (49). A survey of computer-based
empirical models of the ionosphere has been published by Secan (50).
The IRI mentioned previously (see Fig. 21), is a global empirical model that specifies monthly averages
of electron, ion, and neutral temperatures, in addition to electron and ion densities, from about 50 km to
about 2000 km (39). The IRI development is a joint project of URSI and the international Committee of
Space Research (COSPAR), and has proven to be a useful model for scientific research. Another model of
primary interest to workers studying transionospheric propagation effects is the so-called Bent model, a profile
model based upon topside and bottomside sounder data (51). Simplicity is not always important in the age of
sophisticated computers, but the ChingChiu model (52) has found a number of scientific applications in cases
in which detailed ionospheric specification is not paramount.
Aside from global modeling of the ionosphere, there have been attempts to model selected regions of the
world more accurately. During the decade of the 1990s, European scientists affiliated with the COST program
have taken a lead in regional modeling and mapping of the ionosphere (15,16)
Ionospheric Predictions
Ionospheric predictions influence several disciplines, including the prediction of radio system performance, a
matter of some interest in planning as well as ultimate operations. Long-term predictions are generally based
upon predictions of driving parameters such as the sunspot number, the 10.7 cm solar flux, and magnetic
activity indices. Unfortunately, these parameters are not easy to predict. Moreover, the functions relating
these parameters to the ionosphere are imprecise. Therefore, long-term predictions needed for system design
are subject to a considerable amount of uncertainty. To first order the uncertainty in the median value of foF2
for a particular time and location is proportional to the uncertainty in the sunspot number.
In addition to the uncertainty in the mean parameters, we must allow for the fact that ionospheric
parameters have real distributions, and with few exceptions the spread of these distributions is such that errors
about the mean may be a dominant contribution. Short-term ionospheric predictions (or forecasts) generally
refer to departures from the median behavior, the latter being well characterized by running averages of solar
flux and related parameters (e.g., sunspot number). The short-term fluctuations may be specified in terms of
hour-to-hour, day-to-day, and week-to-week variabilities. There are also second-to-second and minute-to-minute
variations, they are generally unpredictable. Compensation for such fluctuations is quite difficult, but may be
accommodated through use of system protocols that enable real-time channel evaluation (RTCE) measures to
be initiated, such as channel sounding or probing. These very short-term forecasts are generally referred to as
nowcasts.
There are four ITU-R documents that are pertinent to the investigation of the ionospheric forecasting
problem. The first deals with the exchange of data forecasts (53), the second outlines various measures for
forecasting of ionospheric parameters (54), the third deals specifically with solar-induced ionospheric effects
(55), and the fourth outlines various real-time channel evaluation schemes (56). These reports should be
consulted.
The distributions of parameters such as foF2, foEs, and hF2 are important, since these parameters depart
significantly from fundamental intuition and from rules set forth by Sidney Chapman in his classic theory.
Distributions of foF2 and foEs are available (57,58), but F 2 -layer height distributions are not directly available.
Ionospheric predictions in the short and intermediate terms provide the most exciting challenge for ionospheric
researchers.
Observational data have shown that TIDs are the ionospheric tracers to a class of atmospheric gravity
waves, and these disturbances are a major contribution to ionospheric variability, especially at F-region heights.
34
IONOSPHERE CHARACTERISTICS
TIDs have a major impact on layer height as well as peak electron density, and possess a variety of scales,
from kilometers to thousands of kilometers. The small- to intermediate-scale TIDs, having wavelengths of less
than a few hundred kilometers and periods of the order of 10 min to 20 min, arise from local sources and have
relatively small amplitudes away from the source region. The large-scale TIDs have sources that are located at
great distances, and there is a strong correlation between this class of disturbances and geomagnetic storms.
Evidence suggests that large-scale TIDs have an impact over global distances and originate within the auroral
zone as a result of atmospheric modifications associated with precipitation and auroral arc formation. A survey
of the effects of TIDs on radiowave systems may be found in a review paper by Hunsucker (59).
The field of ionospheric predictions is undergoing continuous evolution with the introduction of new
scientific methods and instruments, which are providing fresh insight. The requirements for quasi-real-time
products based upon current ionospheric specification has led to increased importance of so-called real-time
ionospheric models. This class of models, in turn, is driven by a hierarchy of solarterrestrial observations,
which enables the analyst to examine the space-weather environment as an integrated complex of phenomena.
This general approach is leading to an improvement in our understanding of ionospheric structure and it
variations, if not better short-term forecasts. In the immediate future, it is anticipated that the primary
ionospheric specification tools will consist of terrestrial sounding systems, including real-time networks of
ionospheric sounders (60). Real-time data services based on these approaches are becoming available (61).
IONOSPHERE CHARACTERISTICS
35
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
K., Davies Ionospheric Radio Propagation, NBS Monograph 80, Washington: USGPO, 1965.
K., Davies Ionospheric Radio Waves, Waltham, MA: Blaisdel, 1969.
K. Davies Ionospheric Radio, IEE Electromagnetic Wave Series 31, London: Peregrinus, Ltd., IEE, 1990.
J. A. Ratcliffe An Introduction to the Ionosphere and the Magnetosphere, London: Cambridge Univ. Press, 1972.
A. Giraud M. Petit Ionospheric Techniques and Phenomena, Dordrecht: Reidel, 1978.
M. C., Kelley The Earths Ionosphere, Plasma Physics and Electrodynamics, San Diego, CA: Academic Press, 1989.
H., Rishbeth Basic physics of the ionosphere, in M. P. M. Hall and L. Barclay, (eds.), Radiowave Propagation IEE
Electromagnetic Series 30, Stevenage, Peregrinus UK, 1991.
A. S. Jursa (scientific ed.), Handbook of Geophysics and the Space Environment, Springfield, VA: Air Force Geophysics
Laboratory, Air Force Systems Command, US Air Force, National Technical Information Service (NTIS), 1985.
T. Tascione Introduction to the Space Environment, Malabar, FL: Orbit, 1988.
R. D. Hunsucker Radio Techniques for Probing the Terrestrial Ionosphere, New York: Springer-Verlag, 1991.
J. M. Goodman HF Communications: Science & Technology, New York: Van Nostrand Reinhold, 1991.
E. E. Johnson Advanced High Frequency Radio Communications, Boston: Artech House, 1997.
L. McNamara The Ionosphere: Communications, Surveillance, and Direction Finding, Malabar, FL: Orbit, Kreiger,
1991.
J. M. Goodman (ed.) Ionospheric Effects Symposia, Springfield, VA: National Technical Information Service. 1975, 1978,
1981, 1984, 1987, 1990, 1993, 1996, 1999,
P. A. Bradley (chair, COST 238 Mgmt.Committee), PRIME: Prediction and retrospective ionospheric modeling over
Europe, Final Report, Commission of European Communities, Chilton, Didcot, Oxfordshire, UK: Rutherford-Appleton
Laboratory, 1999.
R. Hanbaba Improved quality of service in ionospheric telecommunication systems planning and operations, COST 251
Final Report, Warsaw: Space Research Centre, 1999.
S. Chapman The absorption and dissociative or ionizing effect of monochromatic radiation in an atmosphere on a
rotating earth, Proc. Phys. Soc, 43: 26, 1931.
L. M. Muggleton A method for predicting foE at any place and time, Telecomm. J., 42 (7), 413418, 1975.
E. D. Ducharme L. E. Petrie R. Eyfrig A method for predicting the F1 layer critical frequency based upon zurich
smoothed sunspot number, Radio Sci., 8: 837839, 1973.
CCIR, CCIR Atlas of Ionospheric Characteristics, Report 340, General Assembly held in Oslo, Geneva: ITU, 1966.
P. N. Mayaud Derivation, Meaning, and the Use of Geomagnetic Indices, Washington: American Geophysical Union,
1980.
D. F. Muldrew F-layer ionization trough deduced from alouette data, J.Geophys. Res. 70: 2635-2650, 1965.
B. W. Halcrow J. S. Nisbet A Model of F2 peak electron densities in the main trough region of the ionosphere, Radio
Sci., 12: 815-820, 1977.
R. D. Hunsucker R. A. Greenwald, (eds.) Special issue, radio probing of the high latitude ionosphere and atmosphere
new techniques and new results, Radio Sci., 18: 1983.
J. Aarons Equatorial scintillations: A review, IEEE Trans. Antennas Propag., 25: 729736, 1977.
J. D. Whitehead Production and prediction of sporadic E, Rev. Geophys. Space Phys., 8: 65144, 1970.
E. E. Gossard W. H. Hooke Waves in the Atmosphere, Amsterdam: Elsevier Science, 1975.
R. D. Hunsucker Anomalous propagation behavior of radio signals at high latitudes, in H. Soicher (ed.), Propagation
Aspects of Frequency Sharing, Interference and System Diversity, AGARD-CF-332, Springfield, VA: NTIS, 1983.
G. J. Bishop et al. A modern trans-ionospheric propagation sensing system, in Operational Decision Aids for Exploiting
or Mitigating Electromagnetic Propagation Effects, NATO-AGARD-CP-453, UK: Specialised Printing Services, 1989.
Y. I. Feldstein G. N. Starkov Dynamics of auroral belts and polar geomagnetic disturbances, Planet. Space Sci., No. 15,
p. 209, 1967.
M. S. Gussenhoven D. A. Hardy N. Heinemann Systematics of the equatorward diffuse auroral boundary, J. Geophys.
Res., 88: 5692, 1983.
T. A. Chubb G. T. Hicks Observations of the aurora in the far ultraviolet from OGO 4, J. Geophys. Res., 75: 12901311,
1970.
A. P. Mitra Ionospheric Effects of Solar Flares, Dordrecht: Reidel, 1974.
36
IONOSPHERE CHARACTERISTICS
34. B. T. Tsurutani et al. (ed.) Magnetic Storms, Geophysical Monograph 98, Washington: American Geophysical Union,
1997.
35. H. Rishbeth O. K. Garriott Introduction to Ionospheric Physics, New York: Academic Press, 1969.
36. A. Brekke Currents in the auroral zone ionosphere, in The Physical Basis of the Ionosphere in the SolarTerrestrial
System, AGARD-CP-295, London: Tech. Edit. & Reprod., pp. 13.113.9, 1980.
37. L. R. Teters, et al. Estimating the performance of telecommunication systems using the ionospheric transmission
channel, IONCAP users manual, NTIA Report 83-127, PB84-111210, Springfield, VA: U.S. Dept. of Commerce, 1983.
38. ITS, HF Propagation Models: [Online], 2000. Available www: http://elbert.its.bldrdoc.gov/hf.html
39. D. Bilitza International reference ionosphere, NASA, NSSDC 90-22, Greenbelt, MD: World Data Center A (Rockets
and Satellites), 1990.
40. C. M. Rush et al. Ionospheric mappingan update of fof2 coefficients, Telecomm. J., 56: 179182, 1989.
41. P. A. Bradley J. R. Dudeney A simple model of the vertical distribution of electron concentration in the ionosphere, J.
Atmos. Terrest. Phys., 35: 21312146, 1973.
42. M. I. Dick P. L. Bradley The RAL quasi-parabolic model ionosphere Proc. COSTPRIME Workshop on Data Validation
of Ionospheric Models and Maps, COST 238TD (93) 001, 1992, pp. 6783.
43. T. F. Tascione, et al. A technical description of the ionospheric conductivity and electron density profile model (ICED,
Version 196-11), Scott AFB, IL: Hqrtrs Air Weather Service, US Air Force, 1987.
44. T. F. Tascione et al. New ionospheric and magnetospheric specification models, Radio Sci., 23: 211222, 1988; also in
Proc. IES87.
45. D. N. Anderson M. Mendillo B. Herniter A semi-empirical low-latitude ionospheric model Radio Sci., 22: 292, 1987.
46. D. N. Anderson J. M. Forbes M. Codrescu A fully analytic, low latitude ionospheric model, J. Geophys. Res., 94: 1520,
1989.
47. D. Bilitza Solar terrestrial models and application software, Planet. Space. Sci., 40(4): 541579, 1992.
48. M. H. Reilly M. Singh A transionospheric radio propagation model, in J. M. Goodman, (ed.), IES93 Proc. Springfield,
VA: NTIS, 1993.
49. P. H Doherty et al. Validation of PRISM: The climatology, in J. M. Goodman, (ed.), Proc. IES99 Springfield, VA: NTIS,
1999.
50. J. A. Secan A survey of computer-based empirical models of ionospheric electron density, Report NWRA CR-89-11038,
Bellevue, WA: Northwest Research Associates, 1989.
51. R. B. Bent et al. The development of a highly-successful worldwide empirical ionospheric model and its use in
certain aspects of space communications and worldwide total electron content investigations, in J. M. Goodman
(ed.), Effect of the Ionosphere on Space Systems and Communications, IES75, USGPO, Springfield VA: NTIS, 1976,
pp. 1328.
52. B. K. Ching Y. T. Chiu A phenomenological model of global ionospheric electron density in the E, F1 , and F2 -regions,
Atmos. Terrest. Phys., 35: 16151630, 1973.
53. ITU-R, Exchange of information for short-term forecasts and transmission of ionospheric disturbance warnings, ITU-R
Recommendation P.313, Geneva: International Telecommunications Union, 1995.
54. ITU-R, Short-term forecasting of critical frequencies, operational maximum useable frequencies and total electron
content, ITU-R P.888, Geneva: International Telecommunications Union, 1994.
55. ITU-R, Short-term prediction of solar induced variations of operational parameters for ionospheric propagation, ITU-R
P. 727, Geneva: International Telecommunications Union, 1994.
56. CCIR, Real-time channel evaluation of Hf ionospheric radio circuits, Report 889-2, Geneva: International Telecommunications Union, 1990.
57. D. Lucas G. W. Haydon Predicting statistical performance indices for high frequency telecommunication systems,
ITSA-1, Boulder, CO: US Dept. of Commerce, 1966.
58. M. Leftin S. M. Ostrow C. Preston Numerical maps of foEs for solar cycle minimum and maximum, ERL 73-ITS 63,
Boulder CO: US Dept. of Commerce, 1968.
59. R. Hunsucker Atmospheric gravity waves and traveling ionospheric disturbances: thirty years of research, in J. M.
Goodman (ed.), Effect of the Ionosphere on Radiowave Signals and System Performance, IES90, USGPO, Springfield,
VA: NTIS, 1990.
60. I. A. Galkin B. W. Reinisch D. Kitrosser Advances in digisonde networking, in Proc. IES99, J. M. Goodman (ed.),
Springfield, VA: NTIS, 1999.
IONOSPHERE CHARACTERISTICS
37
61. J. M. Goodman J. W. Ballard Dynacast-assisted frequency management for HF communication and broadcasting
systems, in J. M. Goodman (ed.), Proc. IES99, Springfield, VA: NTIS, 1999.
62. NRC, SolarTerrestrial Research for the 1980s, Washington: National Research Council, National Academy Press, 1981.
63. A. Paul Ionospheric Variability, Technical Report 1277, San Diego, CA: Naval Ocean Systems Center, 1989.
64. F. Lied (ed.) High Frequency Radio Communications with Emphasis on Polar Problems, AGARDograph 104, NATO,
Maidenhead, UK: Technivision, 1967.
JOHN M. GOODMAN
Radio Propagation Services, Inc.
494
25
Wavelength (m)
108
50% ionization
hydrogen plasma
20
Highpressure
arcs
Shock
tubes
Attenuation (dB/km)
Laser
plasma
107
101
ii, ee > 1 cm
(4 /3)n Ds < 1
log10 [n (cm3)]
495
Focus
Z pinches
Fusion
reactor
15
106
105
ELF
104
VLF
103
LF
102
MF
101
HF
Day
102
Night
10
104
Alkali
metal
plasma
10
Glow
discharge
Flames
Fusion
experiments
Low
pressure
103
D, > 1 cm
Earth
5 ionosphere
0
2
Earth
plasma sheet
1
2
log10 [T (eV)]
101
100
101
102
Frequency (kHz)
103
Figure 1. Logarithmic plot of approximate magnitudes of some typical laboratory and natural plasmas.
Table 1. The Radio Spectrum as Defined by the International Telecommunications Union, (ITU);
Primary Modes of Propagation, and Effects of the Terrestrial Ionosphere
ITU Designation
Extra low frequency
(ELF)
Very low frequency
(VLF)
Low frequency
(LF)
Medium frequency
(MF)
High frequency
(HF)
Very high frequency
(VHF)
Ultra high frequency
(UHF)
Super high frequency
(SHF)
a
Frequency
Range
30300 Hz
Principal
Propagation Modes
330 kHz
30300 kHz
Same as above
Navigation, standard-frequency
and -time dissemination
Navigation LORAN-Ca
3003000 kHz
AM broadcasting, maritime,
aeronautical communication
Shortwave broadcasting, amateur, fixed services
FM broadcasting, television,
aeronautical communication
Television, radar, navigation,d
aeronautical communication
Radar, space communication
330 MHz
30300 MHz
3003000 MHz
Principal Uses
330 GHz
104
Solar
corona
Solar wind
(1 AU)
102
Submarine communication
496
Winter
400
Summer
F2
400
F2
200
F
F1
200
F1
E
E
12
18
Local time
24
12
18
Local time
24
Perfect reflector
Guide wavelength
Height
EV
2h
image
Source
2 h cos n
Ionosphere
R = 1
EH
E
First-order (TM01) mode
Phase front
EV
Ionosphere
R = 1
Ionosphere z = h
Ground
Perfect reflector
R = +1
Ray direction
z=0
Height
4h
image
R = +1
EH
2h
image
E
Figure 4. Simplified ray geometry for the first-order and secondorder VLFELF modes. The two conducting planes are representative
at the earth surface and the ionosphere.
R = +1
Medium Frequencies
Propagation during the daytime in the MF (300 kHz to 3000
kHz) band is by ground wave, and for frequencies at night
above 500 kHz by sky wave. At geomagnetic latitudes
greater than 55, the auroral ionosphere introduces some
anomalous sky-wave propagation modes (11).
High Frequencies
The ionosphere has the most profound effect on signals in the
HF (3 MHz to 30 MHz) band, making the sky-wave mode the
dominant means of propagation. The ground wave at HF is
sometimes used in the frequency range of 3 MHz to 6 MHz,
especially over sea water, whose conductivity is much greater
than that of ordinary land. At HF wavelengths relatively high
efficiency, gain, and directivity can be achieved in the antenna systems, so directive communications and broadcasting
are realizable. Above 6 MHz, the sky wave is dominant, so
one must really understand ionospheric behavior and phenomenology in order to predict propagation. Since the ionosphere varies with time of day, season, solar activity, and
sunspot cycle, predicting HF propagation over a specific path
can be somewhat complicated. Propagation paths up to
10,000 km are quite common for shortwave (SW) broadcasters, who use antennas with gains of up to 20 dBi (dBi is referenced to an isotropic source) and transmitter powers of 250
kW and higher. With much less reliability, amateur radio operators (hams) sometimes also achieve two-way communications over similar path lengths using antenna gains of 3 dB
to 12 dB and transmitter powers of 5 to 1000 W.
The ionosphere also behaves differently in the equatorial,
midlatitude, auroral, and polar latitudinal regions. Fortunately, several fairly reliable and easy-to-use HF propagation
prediction programs are now available for PCs (IONCAP,
ASAPS, VOACAP, AMBCOM, etc.). The sources of these programs may be found in recent books and articles (5,1214)
and in the amateur radio magazines (QST, CQ, World Radio).
Unfortunately, none of the existing prediction programs gives
very reliable results in the auroral regions. A following section describes the essentials of ionospheric propagation in
considerable detail.
Very High Frequencies
Propagation in the VHF band (30 MHz to 300 MHz) is primarily by line of sight (LOS) to the optical horizon, so if the
antenna patterns direct most of the RF power in the horizontal plane, there are essentially no ionospheric effects. For
earthspace propagation paths, however, the ionosphere can
affect the signal adversely by refraction, diffraction, scattering, or reflection. These effects can be especially important
when the path traverses the equatorial, auroral, and polar
ionosphere. The amplitude, phase, and polarization of the signal may change measurably. These effects will be quantified
in the following section.
Extrahigh Frequencies
At EHF and above, propagation is primarily LOS, and because of the higher frequencies ( f 300 MHz), these signals
are less affected by the ionosphere than lower frequencies. On
earthspace paths that traverse the equatorial and/or high-
497
latitude ionosphere, however, the signal quality can be significantly degraded. These effects will be described below.
To summarize, the radio services most affected by the ionosphere lie in the frequency range of 1 MHz to 150 MHz
[fixed communication services, AM (amplitude modulation)
and SW (shortwave) broadcasting, amateur radio]. To a lesser
degree, services in the 20 kHz to 300 MHz region (mainly
some of the navigation services] suffer some ionospheric perturbation effects.
There is a plethora of radio instrumentation currently deployed globally that operates routinely or on a campaign basis
to measure characteristics of the terrestrial ionosphere. It is
beyond the scope of this article to describe these techniques,
but they have been described in considerable detail in the literature (24).
PHYSICAL PRINCIPLES AND MATHEMATICAL
DESCRIPTION OF ELECTROMAGNETIC
INTERACTION WITH THE IONOSPHERE
Because of the complexity of the terrestrial ionosphere (a
weakly ionized plasma with a superimposed magnetic field in
which electric currents flow), we must utilize the magnetoionic theory to quantify the ionosphere physical parameters.
The most successful formulation of the appropriate magnetoionic theory was derived by Appleton and others in the mid1920s (1517). We can obtain some first-order properties of
the ionosphere by ignoring the magnetic field (18). A simple
dispersion equation for electromagnetic (EM) waves in the
ionosphere is
Ne2
m f 2
(1)
where
refractive index of the ionosphere (real part of the complex refractive index n)
N electron number density of the ionosphere (e/cm3 or
e/m3)
e electronic charge 1.6 1019 C
m mass of the electron 9.1 1031 kg
f frequency of the radio wave in the ionosphere (Hz)
For reflection at vertical incidence, 1 and
N = m f 2 /e2
= 1.24 104 f 2 e/cm3
( f in MHz)
= 1.24 10 f e/m
( f in MHz)
10 2
(2)
fn =
r Ne
(N in cm3 )
= 9 N kHz
= 9 103 N MHz
(N in e/cm3 )
(3)
498
2
c
t=
h
0
dz
dx
1
c T ER sin 0
D
sin 0
=
c
TE + ER
=
c
t=
(4)
h ( f ) =
dz
(5)
1 f n2 / f 2
Since the pulse always travels more slowly in the layer than
in free space, the virtual height of a layer is always greater
than the true height. The true height and virtual height are
related by the integral equation
h ( f ) =
Z max
0
dz
( f, z)
(6)
E
S O
;
;
;;;
;
;;;
D
D
|e|
B 2.80 1010B0
2m 0
(10)
|e|
B 1.76 1011B0
m 0
(11)
Since electrons are much less massive than ions, the electron
gyrofrequency affects the propagation of HF waves in the
ionosphere more than the ion gyrofrequencies. For example,
since B 0.5 104 Wb/m2, the electron gyrofrequency is
1.40 MHz, which falls at the upper end of the medium
wave band.
The Dispersion Relation. Using the recommended URSI (International Union at Radio Science) notation, the magnetoionic dispersion equation for a radio wave in a homogeneous, partially absorbing ionized gas upon which a constant
magnetic field is impressed is given by
n2 = 1
(9)
hob = hv
(7)
(8)
X
YT2
(1 jZ)
2(1 X jZ)
1/2
YT4
2
+ YL
4(1 X jZ)2
where
n complex refractive index ( j)
angular frequency of the exploring wave (rad/s)
(12)
499
Sec (corrected)
800
6.0
5.0
4.0
3.0
2.0
1.0
km
km
00
km
25
00
30
00
00
40
35
600
km
700
500
20
400
300
km
00 km m
00 k
m
18 600 0 k
m
0
1
14 00 k
km
12
0
0
km
10
0
80
m
0k m
60
0k m
0
5
0k
40
km
300
km
200
200
100
m
100 k
12
16
20
24
28
32
36
(13)
j
R=
YL
YT2
1
2 1 X jZ
YT4
1
+ YL2
4 (1 X jZ)2
1/2
(14)
2X (1 X )
2(1 X ) YT2 [YT4 + 4YL2 (1 X )2 ]1/2
(15)
and
R=
Hy
j
=
Hx
YL
1+
X
2
n 1
(16)
40 44 48 52 56 60 70 80
500
ploying two approximations. The quasilongitudinal (QL) approximation applies when the wave is propagating nearly parallel to the geomagnetic field, and the quasitransverse (QT)
approximation applies when the wave propagates in a direction nearly normal to the geomagnetic field. References 21
and 22 contain extended discussions of the QL and QT approximations:
The refractive index n is modified when one introduces collisions between the electrons and heavy particles, and the wave
experiences absorption, which physically is due to the conversion of ordered momentum into random motion of the particles after collision. For each collision, some energy is transferred from the EM wave to the neutral molecules and
appears as thermal energy. We will follow the standard treatment of absorption of radio waves in the ionosphere presented
by Davies (22) and Budden (23).
For the propagation of an EM wave in an unmagnetized
plasma, we can define the absorption index (or coefficient) as
(17)
N
e2
20 mc 2 + 2
(18)
N
dB/km
(2 + 2 )
(19)
We can further simplify Eq. (19) for the VHF case, since
2 2, as
K = 1.15 103
N
dB/km
f2
(20)
N
dB/km
(1 2 X 2 )
2c
(22)
2c
(23)
K=
K=
K=
(21)
Unlike nondeviative absorption, deviative absorption occurs when the wave experiences significant group retardation
and consequently spends a relatively long time in the ab-
K=
2c 1 X
(24)
We should remember that the concepts of deviative and nondeviative absorption are limiting cases, and that as a wave
approaches the reflecting level, ray theory breaks down, so we
must employ full wave theory to obtain a complete description
of the behavior of the wave. Extended discussions of application of the QL and QT approximations to ionospheric absorption may be found in Refs. 2, 5, and 23.
Scattering of Radio Waves in the Ionosphere
The principles of scattering of radio waves in general are discussed in the articles ELECTROMAGNETIC WAVE SCATTERING AND
BACKSCATTER. One can qualitatively describe ionospheric scattering as either strong or weak in terms of the received signal
strength of the scattered signal at the receiving radar antenna. An example of the former is VHFUHF backscatter
echoes received from electron density gradients in the auroral
E region, and an example of the latter is incoherent backscatter received by a VHFUHF radar from the undisturbed E or
F layer.
Another way of classifying scattered echoes is in terms of
their backscatter cross section (using a pulsed radar system)
and their temporal stability. A coherent echo exhibits a statistical correlation of the amplitude and phase from one pulse to
another and emanates from quasideterministic gradients in
electron density, which have correlation times usually greater
than 1 ms, corresponding to a spectral width of the radar echo
of less than 1000 Hz (sometimes less than 100 Hz). It also
has a backscatter cross section 104 to 109 times than that from
an incoherent echo. Other important considerations in the
case of coherent backscatter are the relation between the
scattering-irregularity size relative to the backscatter
sounder free-space wavelength, the mean fractional deviation
in electron density of the scatterer, and the aspect angle between the radar line of sight and the major axis of the irregularity. On the other hand, an incoherent echo arises from random thermal fluctuations in the ionosphere, which have
typical correlation times of 20 s, corresponding to a radar
echo spectral width of 50 kHz.
The physical principles governing coherent and incoherent
scattering from the ionosphere are covered in Refs. 2, 5, and
6, while plasma wave theory is covered in detail in Ref. 8, and
extended descriptions of techniques for studying the ionosphere using coherent- and incoherent-scattering sounders
are given in Refs. 2, 3, 5, and 6.
Scale
magnetic field (m)
10
100
10
Aggregate
of TIDs
(gravitationally
anisotropic)
Phase
scintillation
0.1 0.01
= KHl
Atmospheric scale
height
Plasma turbulence
(magnetically anisotropic)
Plasma waves
(magnetically
anisotropic)
(25)
where
Amplitude
scintillation
Radius of earth
Wandering of normal
to ionosphere
Mutiple normal
to ionosphere
Faraday Rotation
Electron gyroradius
1000 100
Ion gyroradius
Horizontal
scale (km)
501
angle of rotation
K constant associated with each substance
l length of path of light through the substance (m)
H magnetic field intensity (A/m)
Blur on
ionograms
Strong backscattering
and transequatorial
propagation
at VHF
f
=
2c
1
10
X
R
ds
(26)
102
Figure 8. Composite spectrum of ionospheric irregularities as a function of wave number over a large spatial scale. (Courtesy of H. G.
Booker.)
where
Faraday rotation (rad/s)
f wave frequency (Hz)
c 2.998 108 m/s
X kN/f 2
k 80.61
N electron density (e/m3)
and YL, YT are as previously defined.
The integration is between the receiver R, and the satellite S. For VHF frequencies the QL approximation holds and
we can express Eq. (26) as
=
Ionospheric Scintillation
Ionospheric scintillations are fluctuations of amplitude,
phase, and angle of arrival of a VHFUHF signal passing
through irregularities located mainly in the F region. Ionospheric scintillations can have deleterious effects on satellitebased communication and navigation systems. Either extragalactic sources (such as radio stars) or satellite beacon
transmitters may be used as the signal sources for earthobserved studies of ionospheric scintillations, and both geostationary and orbiting satellite beacons have been used.
There is a voluminous body of literature since 1970 describing
the theory, technique, and results of ionospheric scintillation
measurements (25).
K
cf2
f L N ds
(27)
Electric vector
Satellite
antenna
Magnetic
field
Horizontal
502
Impulse
Whistler
f
S
R
f L N ds rad
(28)
(a)
where f L 2.80 10 BL, the electron gyrofrequency corresponding to the longitudinal component of the geomagnetic
field along the ray path.
Details of the application of Faraday rotation theory and
other techniques to deduce ionospheric columnar electron content may be found in Refs. 2, 5, and 6.
10
f (Hz)
6000
4000
2000
0
Time
(b)
Whistlers
1/f (Hz1/2)
0.04
0.03
0.02
0.01
0
8 hop
6 hop
4 hop
2 hop
Lightning
Time
(d)
Figure 10. Sketch of basic manifestations of a whistler and its initiating disturbance. (a) The frequency spectrum. (b) Frequencytime
curve of a typical whistler. (c) Curve of 1/f with time. Initiating
disturbance and multiple hops when the source and receiver are at
the same end of a magnetic line of force. [After Helliwell (25).]
Frequency
where
0.05
Frequency
Whistlers are bursts of EM radiation at VLF that are initiated by lightning discharges and then travel though the ionosphere and magnetosphere in ducts approximately parallel to
geomagnetic lines of force. When translated into sound waves,
whistlers are distinguished by tones of decreasing (or sometimes increasing) frequency, and they may easily be detected
by connecting a suitable antenna to the input of a very sensitive audio amplifier. As a matter of fact, whistlers were first
observed in the last years of the nineteenth century, and were
also heard on the primitive field telephone systems used in
World War I. They have been studied intermittently since
1898, basically as a diagnostic probe of the ionosphere and
magnetosphere (2,25). A graphical representation of whistler
behavior is shown in Fig. 10, and the somewhat rarer nose
whistler behavior is illustrated in Fig. 11. The dispersion relation for whistlers is
f N f L ds
1
D
T=
= 1/2
(29)
1/2
3/2
2c s f ( f L f )
f
tmin
Time
Figure 11. Idealized sketch of the frequency-versus-time characteristics of a nose whistler. [After Davies (5).]
research concerning the effects of the ionosphere on transionospheric propagation and research relating the ionosphere to
the magnetosphere. Most current ionospheric research is related to the interrelationship and coupling between regions
over the entire height region of the terrestrial atmosphere
from the troposphere to the magnetosphere and through interplanetary space to the sun.
There are several areas of ionospheric research that currently seem to be producing exciting new results, and these
areas will probably continue to be emphasized well into the
twenty-first century. These areas include (not necessarily in
order of importance) ionospheric modification by using highpower HF transmissions, ionospheric imaging, coherent radars operating from HF through VHF, and incoherent scattering radars (25). Most of these techniques are employed at
high geomagnetic latitudes as part of the Space Weather Program, but some are also deployed in equatorial regions. We
will briefly describe the essentials of each of these areas of
current ionospheric research emphasis.
Ionospheric Modification
In the late 1960s the availability of military surplus equipment such as very high-voltage and -current power supplies
and HF vacuum tubes capable of many kilowatts of RF output, together with advances in antenna array theory and
practice, induced experimenters to design systems to heat or
otherwise modify the ionosphere. As a result of experiments
performed in the early 1970s at the Platteville, Colorado HF
high-power heating facility (31), some 10 new ionospheric
modification facilities were established and have produced
significant results (see Chap. 14 of Ref. 5). The various modification facilities are listed in Table 2. Other ionospheric modification facilities are located in Russia and Ukraine at Kharkov, Moscow, Zimenki, and Monchegorsk. More information
on the HAARP and other heaters may be obtained on internet
at http://www.haarp.alaska.edu/haarp/airf.html.
Ionospheric Imaging by Radio
For over three decades now, ionospheric scientists have investigated using radio methods to image the ionosphere. Rogers
(32) was probably the first to suggest using the wavefrontreconstruction method for this purpose. Many attempts have
been made to produce holographic images of the ionosphere,
but it has not proven to be a very successful technique
probably because of the difficulty in uniformly illuminating a
large enough horizontal slab of the ionosphere, not using a
sufficient number of receivers, and the inability to achieve
Facility
First
Used
Latitude
Platteville, CO
Arecibo, PR
SURA, Russia
Tromsoe, Norway
HIPAS, Alaska
HAARP,a Alaska
1970
1980
1980
1980
1977
1997
40.2N
18N
56.1N
69.6N
64.9N
62.4N
503
Longitude
Geomag.
Lat.
Transmit
Power
Freq.
Range
(MHz)
Antenna
Gain (dB)
104.7W
67W
46.1E
19.2E
146.8W
145.2W
49
32
71
67
65
62
1.6 MW
800 kW
750 kW
1.5 MW
800 kW
3.6 MWa
2.725
315
4.59
2.58
2.8, 4.5
2.810
18
25
26
28
17
30a
HAARP is currently under construction. Values given are for the completed facility.
504
40
45
50
55
Latitude (deg)
Figure 12. Basic satelliteearth geometry for computer ionospheric tomography, illustrating
the multiple ray paths on which the total electron content is measured.
sp-agency.ca/www/cpus1e.htm, http://thor.ee.cornell.edu/~wes/
CUPRI.
Incoherent Scatter Radars
One of the most powerful earth-based radio methods for
studying the ionosphere is the incoherent scatter radar (ISR)
technique, which has been in use since the early 1960s. ISRs
HF radars
SuperDARN
IS radars
Sondrestrom
EISCAT
Magnetometer chains
Alaska
Canopus
MACCS
SuperDARN
Figure 13. Northern hemisphere map showing area coverage of SuperDARN HF radars,
incoherent scattering radars, and magnetometer chains. (Courtesy of R. A. Greenwald.)
EISCAT
Sondrestrom
Greenland
Magic
Image
505
26. D. Park and D. Carpenter, Very low frequency radio waves in the
magnetosphere, in L. J. Lanzerotti and C. G. Parrs (eds.), Upper
Atmosphere Research in Antarctica, Res. Ser. vol. 29, Monogr. 72,
Washington: American Geophysical Union, 1978.
BIBLIOGRAPHY
506
ELECTROMETERS
48. C. LaHoz (ed.), Special issue: Selected papers from the 6th Int.
EISCAT Workshop, J. Atmos. Terrestr. Phys., 58: 1507, 1996.
ROBERT D. HUNSUCKER
RP Consultants
SPECTRUM CONSIDERATIONS
Medium frequencies and high frequencies are usually dened as the frequency bands from 300 kHz to 3000 kHz
and from 3 MHz to 30 MHz, respectively. However, as far
as ionospheric propagation is concerned, there are no sharp
divisions. The medium-frequency band is dominated by
the amplitude modulated (AM) broadcasting band between
about 500 kHz and about 1700 kHz, which is designed primarily for ground-wave usage. On the other hand, highfrequency systems are designed for long-distance sky wave
propagation and for some 50 years (from about 1925 to
about 1975) provided the primary vehicle for global communications. High frequencies are still used extensively
for communications because of the following advantages:
(1) low cost of terminal equipment, (2) low power requirements, and (3) adequate bandwidths. By contrast, medium
frequencies suffer heavy ionospheric absorption during
daytime and, therefore, there is relatively little co-channel
interference; whereas by night, when the absorption is
small, interference between closely spaced channels is common. High-frequency sky waves suffer from several disadvantages brought about by (1) the temporal and geographical variability of the ionosphere, (2) the large number of
possible propagation paths and the consequent time dispersion of the resulting signal, (3) large and rapid amplitude and phase uctuations, (4) high interference because
of spectrum congestion, and (5) frequency distortion of
wideband signals. An important disadvantage is the occurrence of several types of ionospheric disruptions caused by
solar disturbances. These disruptions can be hemispheric,
such as those on the dayside caused by bursts of solar X
rays (sudden ionospheric disturbances) or conned mostly
to high latitudes (e.g., polar cap disturbances and ionospheric storms that originate in the auroral zones and
spread to populated middle latitudes). Ionospheric storms
are major concerns for high-frequency (HF) users. They occur mostly at high sunspot numbers when the higher critical frequencies help to mitigate their adverse effects (see
Propagation of broadcast transmissions).
In this article we shall discuss the following topics:
PROPAGATION CHARACTERISTICS
The Ionosphere
Sky waves result from radio refraction and reection from
the ionosphere. The ionosphere is usually dened as that
part of the upper atmosphere where sufcient ionization
exists to affect the propagation of radio waves. The ionosphere lies between about 50 km and about 2000 km. The
peak electron density usually occurs in the F region (above
140 km) (Fig. 1). The F region often contains two layers:
the lower F1 layer and the upper F2 layer. Below the F
region is the E region (90 km to 140 km), which contains
the normal E layer and sporadic E. The D region (50 km to
90 km) contains the D layer and the C, or cosmic ray, layer.
Above the F peak is the topside, and above about 2000 km is
the protonosphere. The boundaries between these regions
are not well dened. The alphabetic nomenclature was introduced by Appleton (see Ref. 1, Sec. 1.1), who used the
letter E for electric and F for eld. These letters left room
for the discovery of other layers. Ionospheric electron densities vary by orders of magnitude depending on altitude,
time of day, season, sunspot number, solar disturbances,
and geographical location. It is this variability that renders sky waves so difcult to manage.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
where fN is the plasma frequency in Hz. Ionospheric electron densities are such that typical E layer and F layer
plasma frequencies lie in the range from 0.5 MHz to 30
MHz. Thus, for medium and high frequencies, the refractive index of the ionosphere is less than unity and application of Snells law shows that, on entry into the ionosphere, a wave is refracted away from the vertical. When
the electron density is sufcient, the direction of propagation becomes horizontal and reection occurs. The main
sky wave propagation mechanisms are illustrated in Fig. 2:
absorption, reection, scatter, and penetration. When the
maximum electron density is insufcient, the wave penetrates; this is essential for ground-to-satellite communication. Because ionospheric scatter propagation is considered
elsewhere (see Meteor burst communication), it will not
be discussed in detail in this article.
Absorption
The neutral atmosphere affects the propagation of radio
waves because of electron-neutral collisions that convert
the ordered momentum of the wave into heat and electromagnetic noise. To a rst approximation the loss L, in decibels per kilometer, is
where the + sign refers to the ordinary wave and the sign
to the extraordinary wave, and A is essentially the same for
both waves. The electron gyrofrequency fH is the natural
frequency of rotation of an electron (right-handed) about
the magnetic eld. In the ionosphere fH varies from about
0.8 MHz at the magnetic equator to about 1.6 MHz near the
magnetic poles. Current models of the geomagnetic eld
are available on the Internet (e.g., Ref. see 9). Equation
(3) shows that the absorption of the ordinary wave is less
than that of the extraordinary wave so that an antenna
should be designed to excite as much of the ordinary wave
as possible (see Ref. 1, Sec. 7.6).
The global effects of the geomagnetic eld are 2-fold.
Near the magnetic equator there is an F2-layer anomaly
in which the peak electron densities maximize in the late
afternoon at magnetic latitudes near 15 . Magnetic latitude is dened in terms of the dip angle I by which the
earths magnetic eld dips below the horizontal.
Another way in which the earths eld affects the ionosphere is via the precipitation of magnetospheric charged
particles into the polar caps and the auroral zones (65 to
70 magnetic latitude), where energy is deposited that produces ionospheric storms. In the auroral zones, D-region
electron densities are enhanced producing auroral absorption. Energetic solar protons enter the atmosphere over the
polar caps (latitudes >70 ) and produce intense polar cap
absorption (PCA) that can black out HF signals for several
days.
VERTICAL AND OBLIQUE PROPAGATION
Equivalence of Vertical and Oblique Reection
Equation (1) shows that, with a radio wave incident at an
angle with the vertical on a plane ionosphere, reection
occurs when
and h , when presented graphically, is called a transmission curve. The intersection of a transmission curve with
the ionogram trace gives the virtual heights of reection
of the obliquely traveling signal. When the transmission
curve is tangent to the ionogram, we have the maximum
reected frequency. On frequencies below the maximum
frequency there are two intersections of the transmission
curve and the ionogram trace (for a single layer) showing
that, for a given ground range, there is a low-angle ray and
a high-angle ray. Modications to the plane ionosphere geometry are required for ground ranges over about 200 km
and for the extraordinary ray that determines the maximum frequency, but the principles are essentially the same
(the reader is referred to the aforementioned texts and references therein).
Parabolic Layer Theory
Transmission curves were used extensively for the determination of maximum usable frequencies (MUF) for a
number of years (approximately 1940 to 1980). With the
advent of inexpensive and fast computers an alternative
method has come to the fore. Essentially, the method consists of using ionogram data to approximate the electrondensity prole by an analytic function such as a parabola,
or a quasi-parabola, and calculating the maximum frequency. A parabolic prole is dened by its critical frequency fc , the semithickness ym , and the height of the peak
hm . More complicated ionospheres can be approximated by
two (or more) parabolas or by two parabolas joined by linear segments. The International Reference Ionosphere (5)
also uses a segmented prole. The ground range in each
Ray Tracing
With realistic proles obtained by ionogram inversion,
which have horizontal and vertical structures and which
cannot be represented by simple analytical formulas, there
is recourse to ray tracing. This involves starting with a ray
from a specied transmitting site and with specied angles
of elevation and azimuth, and plotting the ow of energy
step by step until it returns to earth or escapes into space.
This is a complicated procedure and is feasible only with
adequate computers. A comprehensive three-dimensional
ray tracing program has been constructed by Jones and
Stephenson (6) and is available on the Internet (7). In Fig.
3a we see what happens to the ray paths (on a given frequency) as the angle of elevation slowly increases. For low
angles the ground range is long. As the elevation increases,
the ground range decreases until the skip is reached, after
which the range increases rapidly. Eventually penetration
occurs. Figure 3b shows the equivalent triangular paths
for different angles of elevation. The apexes of the equivalent triangles lie on a smooth curve, called a reectrix,
as shown in Fig. 3c. The ray with its apex at the nose of
the reectrix is the skip-ray and its frequency is the MUF.
Rays reected at lower virtual heights are low-angle rays,
and rays reected on the upper side of the reectrix are
high-angle rays. For a given reectrix, the relationship between virtual height and angle of elevation () is given by
Fig. 3d (8).
Figure 3. (a) Real rays in a concentric earth and Chapman ionosphere, f = 8 MHz, fc = 4 MHz,
hmax = 300 km, semithickness = 200 km. (b) Equivalent or virtual paths reected at the apexes of
the ctitious triangular paths. (c) The reectrix as the locus of the virtual reection heights. (d)
The reectrix as a function of virtual height, range, and elevation angle. (Adapted from Croft (8)
by permission of the American Geophysical Union.)
Maximum Frequencies
When the ground range, D, is plotted against the angle
we nd that, for f > fc , there is a minimum range, Ds , called
the skip distance, within which no power is received via
the normal reection process. However, signal power in the
skip zone may result from ground waves, ground scatter,
and scatter from ionospheric irregularities. At the edge of
the skip zone the signal frequency corresponds to the maximum usable frequency for the skip distance, MUF(Ds ). The
MUF for a distance D is related to the critical frequency by
IONOSPHERIC MODELS
Ionospheric models are of two types: (1) empirical models
based on data, and (2) physical models based on production, loss, and movement of plasma by winds and/or electric
elds. For long-term skywave predictions it is customary to
use monthly median models of the main ionospheric characteristics, such as critical frequencies ( fo Ehm E , fEshm F2 ,
fo F1, fo F2) and layer heights (hm E, hm F2) or equivalent M
factors. The diurnal, seasonal, and sunspot number dependencies of the regular E and F1 layers are well behaved and
are represented by analytical expressions (e.g. see Ref. 1,
Chapter 5). However, the most important (F2) layer varies
irregularly, and its characteristics are expressed in the
form of numerical maps of median values. The day-to-day
variability is expressed in terms of upper-decile and lower-
Figure 4. Integrated power in the frequency range 550 kHz to 1600 kHz received in Arviat, North
West Territories, Canada from September 1994 through April 1995. Local meridian (270 E) time
is on the vertical axis. The reception is best near the winter solstice. (Courtesy of J. LaBelle.)
where the frequency, f, is in kHz and is the magnetic latitude at the center of the path for 60 and = 60 for
magnetic latitudes greater than 60 . Thus reception is poor
near the auroral zones. This is illustrated in Fig. 4, which
shows gaps in the reception of signals on days with high
magnetic indexes. This is the result of increased auroral
absorption on the path due to increased D-region electron
content and equatorial motion of the auroral zone. MF signals are sensitive to auroral absorption and above a certain
threshold of disturbance are blacked out.
For eastwest and westeast propagation in low magnetic latitudes (e.g., in Africa), polarization coupling loss
is important. This arises because the electric eld, from a
vertical antenna, is perpendicular to the (horizontal) magnetic eld and excites only an extraordinary wave, which
suffers high D-region absorption [see Eq. (3), with f fH ].
Fading on MF
Nearly all sky waves uctuate, or fade, with time as a result of interference between component echoes, absorption
changes, polarization changes, and so on. To determine a
stations sky wave service area and its interference potential, it is important to know the percentage of time that
a given eld strength is exceeded. For example, elds exceeded for 1% and 10% of the time are about 13 dB and 8 dB,
respectively, higher than the median. In high geomagnetic
latitudes, the corresponding differences are 15 dB and 10
dB (19).
spread sets a limit to the rate of transmission of information because overlapping echoes can result in errors.
Roughly speaking, the maximum rate of transmission (in
binary units per second) is equal to the reciprocal of the
time spread, which is a function of signal frequency, path
length, geographical location, season, sunspot number, and
so on. Ionospheric structures, both vertical and horizontal,
produce echoes with various angles of elevation and azimuth. The interested reader should consult Ref. 4, Chapters 12 through 15, for a detailed discussion of direction
nding.
Fading on HF
Figure 5. Oblique ionogram for the 2370 km, geographic approximately westeast path from
Boulder, Colorado, to Sterling, Virginia, September 1, 1954, 2112 (90 West Meridian Time). Here the
junction frequency (FJF) or basic MUF is the same as the maximum observed frequency (FMOF).
LOF is the lowest observed frequency.
where P is the probability that the mean received signalto-noise ratio exceeds a specied level. The reliability, ,
depends on P increasing with increasing frequency, while
Q decreases with increasing frequency.
Grade of service denes the quality of communication
desired (for example, the percentage of error-free messages
in teletype transmitted or the percentage of satised customers of a given service). Service probability is the fraction of time that a specied grade of service (e.g., signal to
noise) or better is achieved.
Above-the-MUF loss, Lm , accounts for the fact that signals may be received on frequencies above the basic MUF
(e.g., resulting from ionosphere scatter, Es , etc.). The CCIR
(14), Report 252-2, recommends
10
Compatibility is dened as the percentage of time during which a specied criterion of service quality is achieved
at a receiver in the presence of interference, relative to the
value that would be obtained if only noise were present.
Digital System Considerations
Digital systems are particularly affected by ionospheric
dispersion, which produces fading (e.g., by multipath, polarization, etc.). The performance of a digital system is
characterized by its bit error rate, which is the probability that a transmitted binary digit is wrongly detected by
the receiver. On HF, intervals with high error densities alternate with intervals of low error densities. Error bursts
occur when the signal-to-noise ratio temporarily falls below
a critical level (e.g., selective fading). Curves are available
in the ITU (20) Report 197, that gives the duration and
probability of fades as a function of the signal level for specic circuits. In the absence of fading the signal-to-noise
ratio (SNR) is the controlling factor for digital transmission speed. The theoretical error-free channel capacity, C,
is
munications. The pronounced (orders of magnitude) variability of the ionosphere makes sky wave communications
problematic, both as a means of communicating and as a
source of interference. Using HF systems, frequency agility
is essential for continuous operation and, therefore, a set
of suitable assigned frequency channels. Even with such a
frequency allocation, by the appropriate national authority, sky wave propagation is liable to disruption by several solar disturbances (see Propagation of broadcast
transmissions) and synthetic disturbances (e.g., nuclear
explosions and ionospheric modication). For many purposes satellite communications, on gigaHertz frequencies,
have replaced HF for global communications. However, sky
waves will continue to be used well into the foreseeable future.
BIBLIOGRAPHY
1. K. Davies, Ionospheric Radio, London: P. Peregrinus on behalf
of the Inst. Elec. Eng., 1990.
2. J. M. Goodman, HF Communication, Science and Technology,
New York, Van Nostrand Reinhold, 1991.
3. R. D. Hunsucker, Radio Techniques for Probing the Terrestrial
Ionosphere, Berlin: Springer-Verlag, 1991.
4. L. F. McNamara, The Ionosphere: Communications, Surveillance, and Direction Finding, Malibar, FL: Krieger, 1991.
5. D. Bilitza (ed.), International Reference Ionosphere 1990, National Space Science Center, NSSDC 90-20, Greenbelt, MD,
1990.
6. R. M. Jones, J. J. Stephenson, A versatile three-dimensional ray
tracing program for radio waves in the ionosphere, OT Report
75-76, Washington, DC: Government Printing Ofce, 1975.
7. Ionospheric Physics Group, [Online], 1998. Available
http://www.ngdc.noaa.gov/stp/IONO/ionohome.html
8. T. A. Croft, HF radio focussing caused by the electron distribution between ionospheric layers, J. Geophys. Res., 72,
23432355, 1967.
9. National Space Science Data Center, [Online], 1998. Available
http://nssdc.gsfc.nasa.gov/space/model/models/iri.html
10. D. N. Anderson, The development of global, semi-empirical
ionospheric specication models, in J. M. Goodman (ed.), Proc.
Ionospheric Effects Symp., TCI/BR Communications, Alexandria, VA 22314, 1993, pp. 353363.
11. J. C. H. Wang, LF/MF skywave propagation at daytime, IEEE
Trans. Broadcast., 41: 2327, 1995.
12. J. R. Herman, X. A. DeAngelis, Bandwidth expansion effects
on the voltage deviation parameter (Vd) of MF and HF atmospheric radio noise, Radio Sci., 22: 2636, 1987.
13. CCIR, Sky-wave eld-strength prediction method for the
broadcasting service in the frequency range of 150 to 1600
kHz, Recommendation 435-7, 1943, in 1992 CCIR Recommendations, Geneva, Switzerland: Int. Telecommun. Union,
1992.
14. CCIR, Recommendations and Reports of the CCIR 1986,
VI, Propagation in Ionized Media, Geneva, Switzerland: Int.
Telecommun. Union, Report 431-4, 374390, 1986.
15. J. LaBelle et al., The spectrum of LF/MF/HF radio noise
at ground level during substorms, Geophys. Res. Lett., 21:
27492752, 1994.
16. R. D. Hunsucker, B. S. DeLana, High-latitude eld-strength
measurements of standard broadcast skywave transmission
17.
18.
19.
20.
21.
22.
23.
24.
11
KENNETH DAVIES
National Oceanic and
Atmospheric Administration,
Boulder, CO
The vertical component of the electric field generated by a vertical electric dipole of moment Ids, placed
on the ground at the point ( = 0, z = 0) in a flat waveguide with perfectly conducting walls, is described by
Wait (1) as a modal series:
where
= free-space wavelength
k = 2/
120 is the free-space impedance
=
E0 = i(Ids/) e jk /
= Sn sin n
sine of the eigenangle of the nth mode (which may be complex)
Cn = cos n cosine of the eigenangle of the nth mode
n = eigenangle of the nth mode.
Equation (1) describes the field measured within the waveguide, for 0 < z < h. It is a far-field approximation, valid at big enough distances from the source (h and ), where > 300 km can be taken as a
lower limit. See Bannister (2) for a development of field expressions at shorter distances from the source. The
time dependence is ejt , where = 2c/ is the angular frequency.
The other field components, in addition to Ez [Eq. (1)], can be calculated using Maxwells equations, and
they are described by
The modal phase velocity is c/Re[Sn ], and the modal attenuation factor is k Im [Sn ] per unit distance.
The eigenangle n can be interpreted as the incidence angle of the wavefront on the ionosphere, as
indicated in Fig. 1. This interpretation of the eigenangle is closely related to the ray-theory description of
electromagnetic wave propagation in the waveguide, which is an alternative to the modal theory discussed
here. However, the ray-theory treatment of the waveguide is of little consequence in the ELF band, so it is not
discussed further.
The modes generated by the vertical dipole are all transverse magnetic (TM) modes, since the magnetic
field is transverse to the direction of propagation of the wave as it propagates away from the source (and thus
H = 0).
Nickolaenko and Hayakawa (3) calculate the spectrum in the ELF band of the vertical electric field
generated by a vertical dipole, from expressions based on the modal series, and demonstrate the Schumann
resonance, a phenomenon that will shortly be described.
A horizontal electric dipole source of moment Ids, placed at (z = z0 , = 0) parallel to the = 0 direction,
generates a magnetic field whose vertical component is
where
Sm = sin m
= sine of the eigenangle of the mth mode.
Cm = cos
m = cosine of the eigenangle of the mth mode
m = eigenangle of the mth mode
Equation (4) is an approximation valid for 0 z < h, h , and , where > 300 km can be taken
as a lower limit for its validity. The time dependence is ejt . The other components of the field are zero, except
for E and Er .
The modes generated by the horizontal dipole in the broadside direction are transverse electric (TE)
modes, since the electric field is transverse to the direction of propagation of the wave (and thus E = 0). In
other directions, the horizontal dipole generates a mixture of TM and TE modes. The TE waves are weak at
low altitudes (near the ground), because the horizontal electric field is extinguished near the conductive earth.
Since most measurements of ELF fields are made by antennas mounted on the ground, the TE component they
receive is weak. This is possibly the reason that the TE modes receive little attention in the scientific literature.
The modal equations (5) and (6) of the simple flat earth model with perfectly conducting walls determine
the eigenangles of the waveguide:
The lowest TM mode is the 0th, which propagates over the whole frequency range. This mode is also called
the transverse electromagnetic (TEM) mode, because both the electric and the magnetic fields are transverse to
the direction of propagation. The electric field is vertical (Ez component only), and the magnetic field is in the
direction.
The higher modes, namely the TM modes with n = 1, 2,. . . and all the TE modes, propagate above a certain
frequency, called the cutoff frequency, which characterizes each mode. The cutoff frequency of high-order modes
is higher than that of the lower-order modes. Only the first mode is of interest in the ELF range, because the
cutoff frequencies for all the higher modes are above 3 kHz. The cutoff frequency of these first modes (TM and
TE) varies between 1.5 kHz and 4 kHz, depending on the height of the ionosphere.
When extending the waveguide model from the simple flat, perfectly conducting plate model, the following
features may be considered:
The effects of these features of the waveguide on the ELF waves propagating within it are discussed in
the section Extensions of the Waveguide Model below.
Applications
ELF waves propagating in the earthionosphere waveguide undergo very low attenuation (sometimes as
low as 1 dB/Mm to 2 dB/Mm). This allows ELF transmissions to propagate great distances, making this
frequency range suitable for communication networks encompassing the whole globe. Bannister (4) discusses
the attenuation in the ELF band and shows measurements of the attenuation rate at different frequencies.
Another important feature of ELF waves is their good penetration into the earth and sea water. The
skin depth, which characterizes penetration of electromagnetic waves into lossy materials, varies inversely as
the square root of the wave frequency ( 1/ ). Thus, ELF waves have a larger skin depth than waves in
higher frequency bands, and they penetrate better into conducting materials. For example, the skin depth of
waves propagating at 70 Hz in sea water is 30.1 m. The large skin depth is the reason for operating submarine
communication systems in the ELF range. An example of such a system is the United States Navys ELF
Communications System for communication with submarines, operating in the frequency range of 70 Hz to 80
Hz. This system consists of two synchronous transmitters connected to 22.5 km quasi orthogonal antennas.
The major difficulties when communicating in the ELF range are the need for large antennas and the
low data rate. In order to achieve efficient transmission, the physical dimensions of the antennas should be
comparable to the wavelength of the communication frequency, so the long wavelengths inherent in the ELF
range (100 km at 3 kHz up to 100,000 km at 3 Hz) make efficient antenna installation very difficult. This
is a problem mainly in the transmitters, since on the receiving side the antenna preamplifier combination is
limited only by atmospheric noise in the frequency range of interest.
The useful bandwidth in an ELF communication system is limited by the low carrier frequency. Communication systems in the ELF range typically transmit at very low data rates compared to systems utilizing
other frequency bands.
The large antenna dimensions needed for efficient transmission and the inherent low bandwidth are the
reasons that ELF communication systems are few and are usually run by governments for limited military
usage.
Kelly (5) surveys engineering issues related to ELF systems, namely modulation, dispersion, antenna and
system performance, and coverage predictions.
Electrojet Modulation. One scientific application in the ELF band is the modulation of ionospheric
currents, in particular the auroral electrojet. The auroral electrojet is an intense westward current, flowing
in the D and E regions of the ionosphere (the section Ionospheric Reflection below discusses the ionospheric
layers). It flows within the auroral belt, a narrow oval-shaped region encircling the north geomagnetic pole (a
similar belt exists around the south pole).
The intensity of the auroral electrojet is affected by the conductivity of the ionospheric D and E layers.
Changes in the ionospheric conductivity can be induced artificially by heating the ionosphere with radio waves
of appropriate frequency in the HF band, which are efficiently absorbed in the D and E layers. Amplitudemodulated heating of the D and E layers translates into amplitude-modulated conductivity of the medium,
which modulates the intensity of the electrojet. The modulated electrojet reradiates into the earthionosphere
et al. (6) study different modulation methods and polarizawaveguide at the modulation frequency. Villasenor
tions of the heating wave, and Barr and Stubbe (7) discuss the radiation mechanism and evaluate the intensity
of the radiated wave. Papadopoulos et al. (8) discuss the efficiency of the ELF/VLF generation process.
Generation of ELF/VLF signals by modulation of the auroral electrojet has been demonstrated in various
ionospheric heating facilities. Experiments of this kind are discussed by Stubbe et al. (9), Ferraro et al. (10),
Barr et al. (11), Rietveld at el. (12), and Barr et al. (13).
Background Noise
The sources of noise in the ELF range are lightning discharges, man-made noise, and magnetospheric phenomena (such as chorus and hiss); among these, lightning discharges and man-made noise dominate. The low attenuation in this band allows the signals generated by lightning strokes, called atmospherics, or sferics, to travel
great distances around the earth. Thus, a lightning discharge above Africa can be easily detected by a receiver
in North America. It is estimated that on average there are roughly 2000 thunderstorms active around the
earth at any moment, generating an average of 100 to 200 lightning strokes per second. This lightning activity
constitutes the major source of noise in the ELF range. Power-line harmonics are a major source of man-made
noise in the ELF, and their effect is evident throughout the whole frequency band in inhabited areas.
The background power spectrum in the ELF range is generally of the form 1/f [Galejs (14), Chap.
7], where is typically close to 2. Maxwell (15) and Maxwell and Stone (16) show measurements of the
background noise level at various location, and in (15) they discuss the dependence of the background noise
spectrum on latitude. It is shown that the noise spectra correlate well with lightning spectra and propagation
attenuation coefficients. The background noise varies with a period of 24 h because of diurnal variations of
global thunderstorm activity, and the seasonal variation of the noise is related to global thunderstorm activity.
Gustafsson et al. (17) and Smith and Jenkins (18) present background-level measurements recorded in northern
latitudes, and demonstrate that the daytime background level is generally lower than the nighttime level. They
suggest that the reason is enhanced ionospheric absorption during the day.
Fig. 2. Spectrum of the first eight Schumann resonance modes. The basic resonance mode occurs when the wavelength
approximately equals the circumference of the earth.
Consideration of the electrical properties of the earth and ionosphere and calculation of the reflection
coefficients of the waveguide walls
Consideration of the spherical geometry of the earth, which imposes a spherical waveguide instead of a
planar one
Consideration of the geomagnetic field, which causes anisotropic ionospheric reflections.
Reection Coefcients. An accurate model of the earthionosphere waveguide for ELF wave propagation has to take into account the finite conductivities of the two limiting boundaries. The conductivity of
the earth ranges from 10 4 S/m for dry rocky areas to 4 S/m for sea water, and the dielectric constant varies
considerably as well. A list of the conductivity and the dielectric constant values of various types of soil and
water is included in Ref. 34. The earth is usually treated as a uniform sharp boundary, even though this model
does not strictly apply to its changeable surface. The validity of the smooth earth model stems from the small
scale of change of the surface compared to the long wavelengths involved in ELF propagation.
A further extension of the waveguide should take into account the inhomogeneity of the earths surface
conductivity. This extension of the waveguide theory is similar in nature to the theory related to ionospheric
inhomogeneities discussed in the section The Inhomogeneous Ionosphere below.
A realistic model of the ionosphere is more involved, since it depends on the concentration of free electrons
and various positive ions, which in turn depend on complex generation mechanisms. The section Ionospheric
Reflection below discusses the calculation of ionospheric reflection coefficients in more detail.
Another complication of the calculation of the ionospheric reflection coefficients stems from the fact that
the ionospheric boundary is not sharp. In many cases, an effective height is calculated for the waveguide model,
and reflection is treated as though the ionosphere were sharply bounded at this height. Booker and Lefeuvre
(35) calculate the effective ionospheric height for a simple profile of ionospheric ionization density, and show
that it is frequency-dependent. Greifinger and Greifinger (36) present a more complicated reflection model,
which includes two ionospheric reflection layers at different heights.
The reflection coefficients of the earth and the ionosphere affect the eigenangles of the different modes.
These in turn determine the attenuation that each mode undergoes and the relative amplitudes of the modes.
The relation between the eigenangles of the different modes and the attenuation can be appreciated by noting
that if Sn is complex in Equations (1), (2), and (4), then its imaginary part translates into attenuation through
the exponential factor e jkSn . The effect of the reflection coefficients on the eigenangle can be appreciated by
looking at the modal equation, shown here for the TM modes:
where Ri () is the ionospheric reflection coefficient for TM waves, Re () is the earth reflection coefficient for TM
waves, and h() is the frequency-dependent effective reflection height. Both reflection coefficients depend on
the frequency and on the eigenangle n .
Waveguide Geometry. The waveguide geometry affects the various modes [Eqs. (1) and (4)]. It has a
minor effect on the modal equation (9), so the eigenangles are not strongly affected by the flat-earth simplification.
When considering a spherical geometry rather than a parallel-plate geometry, mode coupling is introduced.
This means that TM waves propagating in the waveguide are partially converted to TE waves because of
reflections from the curved walls of the waveguide (and likewise TE waves are converted to TM).
Expressions for the fields in a spherical waveguide are given by Wait (1), using a spherical coordinate
system (r, , ). The earth is modeled as a homogeneous sphere of radius a (= 6370 km), conductivity, g , and
Fig. 3. Geometry of the spherical earthionosphere waveguide with a spherical coordinate system.
dielectric constant g . The lower edge of the ionosphere is located at r = a + h, its conductivity is i , and its
dielectric constant is i . It is assumed that both the earth and the ionosphere are good but not necessarily
perfect conductors. The source is a vertical electric dipole located at ground level at r = a, = 0, = 0, and the
observation point is located on the ground at r = a, , = 0 as in Fig. 3.
The vertical electric field in this configuration is
where
= free-space wavelength
d = a = arc length between the source and the observer
=
120
= free-space impedance
E0 = the field of the source at a distance d on a flat perfectly conducting earth For d/ 1
where
Sn = sin n
= sine of the eigenangle of the nth mode (which may be complex)
Cn = cos n cosine of the eigenangle of the nth mode
n = eigenangle of the nth mode.
Equation (10) describes the field measured within the waveguide, for a < r < a + h. This expression is an
approximation that is valid only if is not near 0 or , which means that the source region and the antipode
are excluded. The time dependence is ejt , where = 2c/ is the angular frequency.
As the radius of the earth a tends to infinity, the flat earth approximation [Eq. (1)] is recovered.
A first-order approximation of the effect of the earth curvature on ELF fields is obtained by multiplying
the field expressions for the parallel-plate waveguide [Eqs. (1) and (4)] by the factor 1 + 2 /12. The coordinate
system used here is cylindrical, as in the section Simplified Waveguide Model above:
Using appropriate approximations, it can be shown that the field consists of two components. One of these
components travels along the short great-circle path connecting the source and the receiver, and the other
travels along the complementary long great-circle path, passing though the antipode [Wait (1)].
An interesting feature of the spherical waveguide is the antipodal concentration that causes the vertical
electric field signals measured at the antipodal point to be stronger than signals measured at a similar distance
away from the source in a plane parallel waveguide. This effect is caused by the addition of many coherent
signal components traveling along great-circle paths between the source and the antipodal point. [The actual
antipodal effect is more complex, due to the daynight terminator (see the subsection DayNight Terminator
Effect below) and other propagation effects in the various paths.] The antipodal concentration and other
antipodal effects are discussed further in the section Antipodal Effects below.
Geomagnetic Field. The static geomagnetic field affects the behavior of the ionospheric boundary of
the waveguide. In the presence of the geomagnetic field, the ionosphere becomes anisotropic and the ionospheric
reflection coefficient depends on the angle between the incident wavefront and the static magnetic field (see
Fig. 4 for the definition of ). This adds a new dependence to the ionospheric reflection coefficient, further
complicating the modal equation:
The direction and intensity of the geomagnetic field depend on the geomagnetic location and, as a result,
the description of the earthionosphere waveguide has to take this location into account.
Dinger et al. (37) show that background noise measured in Norway at frequencies near 3 kHz is not
isotropic. The eastwest component of the measured noise (noise propagating along an eastwest or westeast
path) is 1 dB to 1.5 dB lower than the northsouth component during the day, but increases around local sunset
to a level of about 4 dB above the northsouth component. The difference between noise levels measured in
the two components is explained by the anisotropy of the propagation loss, which is caused by the geomagnetic
field.
Nonreciprocity of ELF Propagation. The geomagnetic field introduces a directional dependence into
the attenuation of ELF waves. This is the reason why ELF waves propagating from west to east suffer higher
attenuation than those propagating from east to west in middle latitudes. This difference between the properties
of the propagation between eastwest and westeast paths is called the nonreciprocity of the propagation.
10
Fig. 4. Geometry of the earthionosphere waveguide with the geomagnetic field present. is the direction of the geomagnetic field relative to the horizontal. is the angle between the incident wave direction and the geomagnetic field.
This effect is analyzed by Barr (38), who shows that the nonreciprocity is most notable for frequencies
above 1 kHz. Barr (39) computes the attenuation factors for east-to-west and west-to-east propagation, and
finds a difference of up to 45 dB/Mm. These results are shown to be in good agreement with measurements.
Antipodal Effects
Some fields measured at the antipodal point of a source are enhanced due to the spherical shape of the
earthionosphere waveguide. The enhancement can be intuitively explained as the coherent addition of many
waves traveling from the source to the antipodal point along great-circle paths. Galejs (14) calculates antipodal
effects on the different field components generated by vertical and horizontal dipole sources. He shows that for a
vertical dipole source the only field component present at the antipode is the vertical component of the electric
field. For a horizontal dipole source the antipodal field contains horizontally directed electric and magnetic
fields, perpendicular to each other, with the electric field parallel to the dipole axis. These results are also
summarized in Ref. 40.
Wait (1) gives a brief description of antipodal effects for a vertical dipole source and shows that waves
traveling through the antipode acquire a /2 phase advance. Wait (41) and Jones (42) analyze the distortion
of ELF pulses after propagation through an antipode. The /2 phase advance of lightning-generated pulses
traveling around the earth is examined.
Measurements of antipodal effects are difficult because of the long distances involved, which dictate
the use of very strong sources. Measurements made on an 82 Hz signal are described in Fraser-Smith and
Bannister (43). The transmitting facility was located in northwestern Russia, and measurements showing
antipodal concentration were made in New Zealand and Antarctica. Similar results, in the VLF band, are
reported by Rogerson (44). In this latter case the source was a U.S. Navy transmitter in Hawaii, and the
measurements were taken by an aircraft in the vicinity of the antipodal point located in southern Africa.
Ionospheric Reection
This section describes the mechanism of reflection of ELF waves from the ionosphere. The ionosphere is
conductive because it contains free electrons and positive ions, which are generated largely by solar ultraviolet
11
Fig. 5. Typical electron densities at 44.6 N, 2.2 E at noon and midnight on March 21, for solar minimum and maximum
conditions. The electron density determines the behavior of the ionospheric wall of the earthionosphere waveguide.
Adapted from A. D. Richmond, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, CRC Press, 1995, with
permission (CRC Press home page http://www.crcpress.com).
(UV) and X radiation and by the precipitation of energetic charged particles from the magnetosphere. A
discussion of the ionization generation mechanism and the various positive ions involved can be found in
Richmond (45) and in other literature on the subject [Rees (46), Kato (47), Akasofu and Chapman (48), Ratcliffe
(49,50), Rishbeth and Garriot (51), Whitten and Poppoff (52), and Rawer (53)].
The amount of free charge in the ionosphere is determined by the equilibrium that is reached between
the generation mechanisms and the recombination of the free electrons with the positive ions. The density
of neutral particles in the ionosphere decreases exponentially with increasing altitude, which means that
very little generation occurs at the high altitudes (above 500 km), where the particle density is low. The
energetic particles precipitating from the magnetosphere gradually lose their energy as they penetrate the
ionosphere, generating free charge as they travel downward. The solar radiation penetrating the ionosphere
is likewise gradually absorbed while generating free charge in the ionosphere. The bottom of the ionosphere
(around 50 km) is marked by very little generation of charge, because the ionizing radiation has lost most of its
energy at higher altitudes. The density of free charge in the ionosphere is maximal at middle heights, where
the generating mechanisms are strong and the density of particles is sufficient. The free-electron density is
maximal at an altitude of about 250 km above ground, with a few local maxima of density at lower altitudes.
These local maxima constitute ionospheric layers with enhanced electron density, as seen in Fig. 5.
The layer about the altitude of 100 km is called the E layer, the layer at 170 km, which is sometimes
absent at night, is the F1 layer, and the top layer at 250 km is the F2 or the F layer. The lowest layer, at about
70 km, is the D layer, which is almost always absent at night. A description of the structure and properties of
the ionosphere is presented by Hargreaves (54).
The electron density height profile in the ionosphere differs greatly between day and night conditions,
because of the different ionization generation rates driven by solar radiation. The electron density is also
affected by solar activity, with the ionization being lowest during periods of low solar activity, as seen in Fig. 5.
The ionosphere is a plasma, and its conductivity can be calculated using plasma theory, as described by
Rishbeth and Garriot (51), by Ratcliffe (49), and in other literature on the subject. The ionospheric plasma
parameters (i.e., the types of particles and their concentrations) are altitude-dependent, so the conductivity of
12
Fig. 6. Typical noontime parallel (0 ), Pedersen (1 ), and Hall (2 ) conductivities at 44.6 N, 2.2 E for solar minimum
conditions on March 21. The different types of conductivity characterize the behavior of the ionosphere as a reflector of ELF
waves. Adapted from A. D. Richmond, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, CRC Press, 1995,
with permission (CRC Press Home Page http://www.crcpress.com).
the ionosphere also varies with altitude. In addition, the conductivity varies in the lateral plane, with the most
notable variation occurring at the daynight terminator, which marks the dawn and dusk boundaries. Local
disturbances in the ionosphere also cause lateral variations of the conductivity; these variations are discussed
in the section The Inhomogeneous Ionosphere below.
Due to the geomagnetic field, the ionosphere is an anisotropic conductor. The three most commonly used
conductivities apply to specific directions relative to the geomagnetic field, as follows:
The Parallel conductivity 0 relates the current flowing along the static magnetic field to an electric field
component in the same direction. This conductivity is not affected by the presence of the static magnetic
field.
The Pederson conductivity 1 corresponds to current flowing perpendicular to the static magnetic field when
an electric field is applied in the same direction.
The Hall conductivity 2 corresponds to current flowing in a direction perpendicular to both the magnetic
field and the electric field, when the electric field is perpendicular to the magnetic field.
Figure 6 shows typical midlatitude vertical profiles of the daytime conductivity components for low
solar activity. At altitudes above 80 km the parallel conductivity is much larger than the Pederson and Hall
conductivities.
The reflection coefficients of the ionosphere depend on (a) the conductivity profile of the ionosphere, (b)
the local geomagnetic field, (c) the direction of incidence, and (d) the polarization of the incident wave. Due to
the complexity of the reflection coefficients, they are usually calculated numerically or empirically. Simplified
empirical expressions for the reflection coefficients in day and night conditions are given in Volland (55) for
a simplified model characterized by an exponential electron density and an exponential collision rate in the
lower layers of the ionosphere. The effects of the geomagnetic field are neglected. The reflection coefficient for
TM waves under these simplified conditions is
where bi and ci vary exponentially with frequency and is the incidence angle.
13
Harth (34) calculates the ionospheric reflection coefficient at 3 kHz using a thin-layer method (see the
final section for a discussion of the thin-layer method) with similar assumptions to Vollands, and the results
are presented graphically. Wait and Spies (56) present graphs describing the reflection coefficient for an
exponential electron density profile under various conditions. They show the effects of the magnetic field, angle
of incidence, frequency, reflecting height, and ground conductivity on the reflection coefficient. An experiment
relating electron density profiles to VLF reflection coefficients is described by Mambo et al. (57).
A different approach to the calculation of the reflection coefficients is taken by Budden (58), who presents
a plasma physics approach to understanding the ionosphere. The calculation of the refractive index and
the characteristic polarization of waves traveling through the ionosphere is the basis of this approach. The
admittance matrix of the ionosphere is calculated, which relates the horizontal components of the electric and
magnetic field, and the reflection coefficients of the ionosphere are then derived from the matrix expressions.
Two types of ionospheric boundaries are considered: a sharp boundary and a slowly varying one [Wentzel
KramersBrillouin (WKB) model]. A similar approach is taken by Galejs (14), where the modal equations are
presented in terms of the wave impedance rather than the reflection coefficient.
The ionosphere can usually be modeled as a sharp boundary for frequencies in the ELF range, since the
wavelengths are large compared to the scales of length over which the ionospheric properties change. As a
result, a simplified model of the ionosphere with a sharp bottom boundary is often sufficient.
14
Fig. 7. Amplitude of the vertical electric field in a propagation path inclined 45 from the terminator, relative to the
unperturbed field in the models of the smooth and sharp terminators. The day night terminator is the main source
of lateral variation in the earthionosphere waveguide. Adapted from A. P. Nikolaenko, in Radiophysics and Quantum
Electronics, Vol. 29, No. 1, Plenum Publishing Corporation, 1986, with permission.
Ionospheric Disturbances.
Varieties of Disturbances. Three major varieties of ionospheric disturbances are: (1) sudden ionospheric
disturbances (SIDs), (2) magnetic storms, and (3) solar proton events (SPEs), also called proton storms (51).
They can be defined as follows:
Sudden ionospheric disturbances are related to solar flares, electromagnetic phenomena that enhance the
electron density of the D region by up to a factor of ten.
Magnetic storms are caused by enhancements of the solar wind, a stream of particles traveling away from
the sun. An increase in the particle concentration of the solar wind, accompanied by an increase in their
velocity, causes ionospheric modifications where it reaches the magnetosphere, which result in magnetic
storms. The solar wind enhancements are usually associated with the solar events known as coronal mass
ejections or CMEs.
Solar proton events are caused by energetic streams of solar protons (accompanied by electrons). On arrival
at the earth the protons are deviated by the geomagnetic field, and they impinge on the polar caps, so solar
proton events are related to polar cap absorption (PCA) events.
Other ionospheric disturbances include the effect of sporadic E layers and small-scale ionospheric irregularities. A summary of ionospheric disturbances and their effect on radio propagation is given by Hargreaves
(54).
General Considerations. The theoretical analysis of ionospheric inhomogeneities often involves calculations of the electromagnetic field scattered from the inhomogeneities. Pappert (64) calculates the field scattered
from a cylindrically symmetric ionospheric disturbance and compares the expected effect of disturbed polar-cap
15
boundaries with measurements. The cylindrically symmetric ionospheric disturbance model is also useful in
describing small-scale ionospheric irregularities.
The effect of local ionospheric disturbances on ELF propagation is analyzed by Nickolaenko (65,66) using
a perturbation theory. The disturbances analyzed are (1) local variations of the electrical properties of the
ionosphere and (2) changes in the effective ionospheric reflection height. The analysis shows that the effects
of these two types of disturbances on ELF propagation are similar and that the effect of local disturbances is
greatest when the irregularity is situated above the source or the observer.
Random distortions in ionospheric height and fluctuations of electrical properties cause fluctuations of
ELF waves propagating in the earthionosphere waveguide; these are analyzed by Kostigov (67) using a
perturbation method.
Specic Effects. Ionospheric inhomogeneous phenomena affect ELF waves traveling between the ionosphere and the earth. Bannister (4) reviews measurements of various ELF propagation phenomena related to
ionospheric irregularities, some of which are mentioned below.
The effects of sudden ionospheric disturbances caused by solar flares on ELF propagation are analyzed
by Field (68), and the resulting modal attenuation coefficients are compared with measurements. The X-ray
radiation associated with solar flares causes a lowering of the D layer, which reduces the propagation loss in
the ELF range relative to nondisturbed conditions.
Precipitation of energetic electrons in the several days following magnetic storms may cause enhanced
loss in the ELF range at night. Measurements showing this effect are described by Davis (69) and Bannister (4),
but the theoretical explanation is incomplete. Three different groups analyzed a magnetic storm that occurred
in December 1971 using satellite measurements, and all of them measured enhanced electron precipitation
and enhanced levels of ELF radiation [Parady and Cahill (70), Vakulov et al. (71), and Kovner et al. (72)]. These
results suggest that the relationship between the electron precipitation and the ELF fields is connected with
the location of the magnetopause, that is, the outer termination of the geomagnetic field.
Occasional nighttime decreases in ELF signal levels may be explained by the presence of sporadic E
layers [Bannister (4,73)]. The measurements described by Bannister include ELF anomalies at two different
locations, appearing 2 h to 4 h apart, which may be the result of a traveling sporadic E layer. Further analysis
of the measurements from these two locations yields an estimate of the velocity and path of the sporadic E
layer [Pappert (74)].
Theoretical analyses of the effect of sporadic E layers on ELF propagation are given by Pappert (75) and
by Pappert and Moler (76). Their results indicate that a sporadic E layer 1 Mm by 1 Mm may cause 6 dB to 8
dB attenuation in the ELF range. Patches 1 Mm by 0.5 Mm can account for fades in the 3 dB to 4 dB range. The
deepest fades require the center of the disturbance to be close either to the source or to the receiver, but large
patches cause significant attenuation at distances of up to 10 Mm from the direct propagation path between
the source and receiver.
The main effect of a SPE is an increase in the attenuation in the ELF range for paths passing near the
boundary of the polar cap [Katan and Bannister (77), Field et al. (78) and Fraser-Smith and Helliwell (79)].
The additional attenuation may be caused by lateral refraction, which bends the signal path away from the
polar-cap boundary and into the central cap zone. Field et al. (78) predict enhanced field levels inside the
polar-cap boundary, but this was not verified by measurement.
The expected effects of nuclear events on ELF propagation via ionospheric changes is surveyed by Bannister (4).
16
calculations are often done numerically, incorporating information about various ionospheric and ground parameters.
The two main computational approaches presented here, namely (a) the full-wave approach and (b) the
thin-layer approach, are suitable for a horizontally stratified ionosphere with no lateral variation. Calculation
methods for laterally varying ionospheres are also surveyed.
A noniterative approximate method for calculating eigenvalues of the ELF modes, based on a two-layer
ionospheric model, is presented by Greifinger and Greifinger (36) and elaborated by them in Ref. 80. The
physical significance of a reflecting layer for ELF waves within the ionosphere is discussed in those papers.
The Full-Wave Approach. The full-wave approach involves the numerical integration of four coupled
first-order complex differential equations containing an eigenvalue parameter [Greifinger and Greifinger (80)].
The calculated quantities may be (a) the components of the electromagnetic field, (b) ratios of field components,
which may represent impedances, and (c) reflection coefficients. The full-wave integration starts from the top
of the ionosphere, where propagation is assumed to be upward only. An initial value of propagation parameters
is assumed at this height. Integration is performed downward along a vertical path, and boundary conditions
are checked at the surface of the earth. Inconsistencies of the boundary conditions at the surface of the earth
lead to corrections of the initial value (of propagation parameters at the top of the ionosphere), and the whole
process repeats with new initial values, until the boundary conditions are matched to some chosen accuracy.
A full-wave calculation of the ionospheric reflection coefficients is described by Pappert and Moler (81).
The ionosphere is assumed to vary slowly along the lateral propagation path, and the WKB approximation is
used.
Field (82) uses a full-wave method to calculate the ionospheric wave admittance for normal and disturbed
ionospheric conditions, and calculates the fields within the waveguide.
Aksenov and Nazarova (83) use a full-wave method to calculate the dependence of ionospheric reflection
and transmission coefficients on the angle of incidence, frequency, and geomagnetic field for ionospheric models
of the day and night.
The Thin-Layer Approach. The thin-layer approach is similar in principle to the full-wave approach,
but the ionosphere is taken to be a collection of thin homogeneous layers, rather then a continuum. The process
is started with some initial values of the parameters at the top boundary of the ionosphere, and then the
boundary conditions are matched between adjacent layers, going downward until a lower boundary is reached.
The thin-layer method is iterative in a similar way to the full-wave method: the boundary conditions at the
lower boundary are used to correct the initial values, and the calculation process is repeated until convergence
is achieved to a desired accuracy.
Galejs (84) presents a thin-layer method where the surface impedance computed at a lower boundary is
used to correct the initial conditions. The field components are calculated for each layer, and from these the
wave impedances are derived. This procedure is used to calculate the frequency-dependent attenuation rate
for day and night conditions and to examine the effect of the anisotropy of the ionosphere.
A similar approach is presented by Altman and Cory (85,86), where the ionosphere is treated as a
collection of discrete layers and an iterative process is used to calculate ELF propagation parameters, each
iteration starting from the top of the ionosphere. This method is presented as a derivation from the thin-film
method used in optics.
Another thin-layer method is presented by Nagano et al. (87), where the calculated quantities are the
components of the field. Nagano et al. (87) compare the width of the layers required for their method with the
step size required in various full-wave methods and show that their thin-layer method produces comparable
results to full-wave methods, with much bigger step sizes. The complexity (computer time) of the two methods
is also compared, and it is shown that the thin-layer method is more efficient.
Lateral Variations in the Waveguide. Lateral variations in the earthionosphere waveguide may be
treated by the WKB approach, which assumes slow variation in the waveguide. However, this approach is of
17
little use in the ELF band, because the wavelengths involved are very large, with the result that in many cases
the ionospheric or ground variations are not slow compared to the wavelength.
An important approach to lateral variations in the waveguide is based on an integral equation describing
the lateral propagation [Field and Joiner (88)]. The integral equation is solved numerically in terms of the sine of
the eigenangle. This approach is based on the simplified assumptions that vertical variations in the ionosphere
are decoupled from lateral variations and that lateral variations are much slower than vertical variations. A
program that predicts vertical ELF fields in an inhomogeneous waveguide is described by Ferguson et al. (89).
This program uses the moment method to solve the integral equation numerically.
Another approach to the laterally varying waveguide is based on ray theory, where vertical inhomogeneities are assumed to be uncoupled from the lateral variations. The vertical inhomogeneities are usually
treated in a full-wave approach, and the lateral variations are treated using ray theory. Field et al. (78) use the
full-waveray method to analyze the effects of polar-cap absorption events on ELF propagation.
A calculation of the electric field, given the eigenvalues of the propagation, is described by Shellman (90).
The method is based on the knowledge of the sine of the waveguide eigenangles at elements of a mesh-point
array that extends around the earth. The lateral variation of the waveguide is described by the eigenvalues at
different locations.
Acknowledgments
BIBLIOGRAPHY
1. J. R. Wait Electromagnetic Waves in Stratified Media, revised ed. 1970. New York: Pergamon Press.
2. P. R. Bannister Simplified formulas for ELF propagation at shorter distances , Radio Sci. 21 (3): 529537, 1986.
3. A. P. Nickolaenko M. Hayakawa Natural electromagnetic pulses in the ELF range , Geophys. Res. Lett. 25 (16): 3103
3106, 1998.
4. P. R. Bannister ELF propagation update , IEEE J. Oceanic Eng. OE-9: 179188, 1984.
5. F. J. Kelly ELF/VLF/LF propagation and system design, Technical Report NRL 9028, Naval Research Laboratory, June
1987.
6. J. Villasennor
et al. Comparison of ELF/VLF generation modes in the ionosphere by the HIPAS heater array , Radio
Sci., 31 (1): 211226, 1996.
7. R. Barr P. Stubbe. ELF and VLF radiation from the polar electrojet antenna, Radio Sci., 19 (4): 11111122, 1984.
8. K. Papadopoulos et al. On the efficiency of ionospheric ELF generation, Radio Sci., 25 (6): 13111320, 1990.
9. P. Stubbe et al. Ionospheric modification experiments with the Troms Heating Facility, J. Atm. Terrestrial Phys., 47
(12): 11511163, 1985.
10. A. J. Ferraro et al. Measurements of extremely low frequency signals from modulation of the polar electrojet above
Fairbanks, Alaska, IEEE Trans. Antennas Propag., 37: 802805, 1989.
11. R. Barr et al. ELF and VLF signals radiated by the polar electrojet antenna: Experimental results, J. Geophys. Res.,
91 (A4): 44514459, 1986.
12. M. T. Rietveld et al. Ionospheric heater beam scanning: A new technique for ELF studies of the auroral ionosphere,
Radio Sci., 19 (4): 10691077, 1984.
13. R. Barr et al. Ionospheric heater bean scanning: A mobile source of ELF radiation. Radio Sci., 22 (6): 10731083,
1987.
14. J. Galejs Terrestrial Propagation of Long Electromagnetic Waves, New York: Pergamon, 1972.
15. E. L. Maxwell Atmospheric noise from 20 Hz to 30 kHz, Radio Sci., 2 (6): 637644, 1967.
18
16. E. L. Maxwell D. L. Stone Natural noise fields from 1 cps to 100 kc, IEEE Trans. Antennas Propag. AP-11 (3): 339343,
1963.
17. G. Gustafsson A. Egeland J. Aarons Audio-frequency electromagnetic radiation in the auroral zone, J. Geophys. Res. 65
(9): 27492758, 1960.
18. A. J. Smith P. J. Jenkins A survey of natural electromagnetic noise in the frequency range f = 110 kHz at Halley
station, Antarctica: 1. Radio atmospherics from lightning, J. Atm. SolarTerrestrial Phys., 60 (2): 263277, 1998.
19. J. R. Wait Earthionosphere cavity resonances and the propagation of ELF radio waves, Radio Sci., 69D (8): 10571070,
1965.
20. D. L. Jones Numerical computations of terrestrial ELF electromagnetic wave fields in the frequency domain, Radio
Sci. 5 (5): 803809, 1970.
21. D. B. Large J. R. Wait Theory of electromagnetic coupling phenomena in the earthionosphere cavity, J. Geophys. Res.
73 (13): 43354362, 1968.
22. M. Balser C. A. Wagner Observations of earthionosphere cavity resonances, Nature, 188: 638641, 1960.
23. M. Balser C. A. Wagner Diurnal power variations of the earthionosphere cavity modes and their relationship to
worldwide thunderstorm activity, J. Geophys. Res. 67 (2): 619625, 1962.
24. A. P. Nickolaenko L. M. Rabinowicz Study of the annual changes of global lightning distribution and frequency variations of the first Schumann resonance mode, J. Atm. Terrestrial Phys. 57 (11): 13451348, 1995.
25. A. P. Nickolaenko M. Hayakawa Y. Hobara Temporal variations of the global lightning activity deduced from the
Schumann resonance data, J. Atm. Terrestrial Phys. 58 (15): 16991709, 1996.
26. T. Ogawa Y. Tanaka M. Yasuhara Schumann resonances and worldwide thunderstorm activity, in S. C. Coroniti and J.
Hughes, (ed.), Planetary Electrodynamics, Vol. 2, New York: Gordon and Breach, 1969, Chap. V-7.
27. S. J. Heckman E. Williams B. Boldi Total global lightning inferred from Schumann resonance measurements, J. Geophys.
Res. 103 (D24): 31, 77531, 779, 1998.
28. C. Polk Relation of ELF noise and Schumann resonances to thunderstorm activity, in S. C. Coroniti and J. Hughes (ed.),
Planetary Electrodynamics, Vol. 2, New York: Gordon and Breach 1969, chap. V-6. pp. 5863.
29. J. Galejs Frequency variations of Schumann resonances, J. Geophys. Res. 75 (16): 32373251, 1970.
30. D. D. Sentman B. J. Fraser Simultaneous observations of Schumann resonances in California and Australia: Evidence
for intensity modulation by the local height of the D region, J. Geophys. Res. 96 (A9): 15,97315,984, 1991.
31. K. Sao et al. Experimental investigations of Schumann resonance frequencies, J. Atm. Terrestrial Phys. 35: 20472053,
1973.
32. E. R. Williams The Schumann resonance: A global tropical thermometer, Science 256: 11841187, 1992.
33. A. P. Nickolaenko Modern aspects of Schumann resonance studies, J. Atm. SolarTerrestrial Phys., 59 (7): 805816,
1997.
34. W. Harth Theory of low frequency wave propagation, in H. Volland (ed.), CRC Handbook of Atmospherics, Vol. II, CRC
Press, Boca Raton, FL, 1982, pp. 133202.
35. H. G. Booker F. Lefeuvre The relation between ionospheric profiles and ELF propagation in the earthionosphere
transmission line, J. Atm. Terrestrial Phys., 39: 12771292, 1977.
36. C. Greifinger P. Greifinger Approximate method for determining ELF eigenvalues in the earthionosphere waveguide,
Radio Sci., 13 (5): 831837, 1978.
37. R. J. Dinger W. D. Meyers J. R. Davis Experimental investigation of ambient electromagnetic noise from 1.0 to 4.0 kHz
in Italy and Norway, Radio Sci., 17 (1): 285302, 1982.
38. R. Barr Some new features of ELF attenuation, J. Atm. Terrestrial Phys. 34: 411420, 1972.
39. R. Barr The propagation of ELF and VLF radio waves beneath an inhomogeneous anisotropic ionosphere, J. Atm.
Terrestrial Phys. 33: 343353, 1971.
40. J. Galejs Near fields and antipodal fields in the terrestrial earthionosphere waveguide, Proc. IEE 116 (7): 11501158,
1969.
41. J. R. Wait Distortion of an ELF pulses after propagation through an antipode, J. Geophys. Res. 74 (11):29822986, 1969.
42. D. L. Jones Propagation of ELF pulses in the earthionosphere cavity and application to slow tail atmospherics, Radio
Sci. 5 (8,9): 11531162, 1970.
43. A. C. Fraser-Smith P. R. Bannister Reception of ELF signals at antipodal distances, Radio Sci. 33 (1): 8388, 1998.
44. J. E. Rogerson Airborne field strength measurements in the region of the NPM antipode, Radio Sci. 2 (6): 581587,
1967.
19
45. A. D. Richmond The ionospheric wind dynamo: Effects of its coupling with different atmospheric regions, in R. M.
Johnson and T. L. Killeen (eds.), The Upper Mesosphere and Lower Thermosphere, Washington: American Geophysical
Union, 1995.
46. M. H. Rees. Physics and Chemistry of the Upper Atmosphere, Cambridge University Press, 1989.
47. S. Kato Dynamics of the Upper Atmosphere, Dordrecht: Reidel, 1980.
48. S. I. Akasofu S. Chapman SolarTerrestrial Physics, Oxford: Clarendon, 1972.
49. J. A. Ratcliffe An Introduction to the Ionosphere and Magnetosphere, Cambridge University Press, 1972.
50. J. A. Ratcliffe Sun, Earth and Radio, McGraw-Hill, 1970.
51. H. Rishbeth O. K. Garriot Introduction to Ionospheric Physics, Academic Press, 1969.
52. R. C. Whitten I. G. Poppoff Physics of the Lower Ionosphere, Engelwood Cliffs, NJ: Prentice-Hall, 1965.
53. K. Rawer The Ionosphere: Its Significance for Geophysics and Radio Communications, New York: Ungar, 1957.
54. J. K. Hargreaves The SolarTerrestrial Environment, Cambridge Atmospheric and Space Science Series, Cambridge
University Press, 1992.
55. H. Volland Longwave sferics propagation within the atmospheric waveguide, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, Vol. II, Boca Raton, FL: CRC Press, 1995, Chap. 3, pp. 6593.
56. J. R. Wait K. P. Spies Characteristics of the earthionosphere waveguide for VLF radio waves, Technical Note 300,
National Bureau of Standards, 1964.
57. M. Mambo, et al. An improved method of estimating electron density profiles in the lower ionosphere with VLF reflection
coefficients, Electron. Commun. Japan, 67-B (7): 2837, 1984.
58. K. G. Budden The Propagation of Radio Waves, Cambridge University Press, 1985.
59. H. G. Hughes R. J. Gallenberger Propagation of extremely low-frequency (ELF) atmospherics over a mixed daynight
path, J. Atm. Terrestrial Phys. 36: 16431661, 1974.
60. J. R. Davis ELF propagation irregularities on northern and mid-latitude paths, in J. A. Holtet, (ed.), ELFVLF Radio
Wave Propagation Proc. NATO Advanced Study Institute, April 1727, 1974, Dordrecht: D. Reidel, 1974, pp. 263277.
61. V. G. Bezrodnyi et al. Reflection of superlong waves from the terminator in the earthionosphere waveguide, Radiophys.
Quantum Electron., 21 (11): 11001114, 1978.
62. E. C. Field R. G. Joiner Effects of lateral ionospheric gradients on ELF propagation, Radio Sci., 17 (3): 693700, 1982.
63. A. P. Nikolaenko Scattering of ELF radio waves on global inhomogeneities of the earthionosphere cavity, Radiophys.
Quantum Electron., 29 (1): 2632, 1986.
64. R. A. Pappert ELF scattering in the earthionosphere waveguide, Radio Sci., 24 (5): 629639, 1989.
65. A. P. Nickolaenko ELF radio wave propagation in a locally nonuniform earthionosphere cavity, Radio Sci., 29 (5):
11871199, 1994.
66. A. P. Nikolaenko Effects of a local inhomogeneity in the ionosphere on the propagation of ELF radio waves, Radiophys.
Quantum Electron., 27 (10): 856864, 1984.
67. K. I. Kostigov Theory of fluctuations of super-long radio waves caused by random deformations of the boundaries of the
earth surfaceionosphere waveguide channel, Radio Eng. Electron. Phys., 19 (8): 123126, 1974.
68. E. C. Field VLF and ELF propagation during sudden ionospheric disturbances, J. Geophys. Res., 75 (10): 927933, 1970.
69. J. R. Davis Localized nighttime D-region disturbances and ELF propagation, J. Atm. Terrestrial Phys., 38: 13091317,
1976.
70. B. Parady L. J. Cahill, Jr. ELF observations during the December 1971 storm, J. Geophys. Res., 78 (22): 47654770,
1973.
71. P. V. Vakulov et al. Spilling of charged particles from the radiation belts and variations in the intensity of the ELF
radiation in the recovery phase of the magnetic storm of December 1617, 1971, Cosmic Res., 12 (5): 645652, 1974.
72. M. S. Kovner et al. High-energy electron flux and ELF radiation flux at ionospheric heights during the magnetic storm
of December 16, 1971, Cosmic Res., 15 (3): 374379, 1977.
73. P. R. Bannister Localized ELF nocturnal propagation anomalies, Radio Sci., 17 (3): 627634, 1982.
74. R. A. Pappert Calculated effects of traveling sporadic E on nocturnal ELF propagation: Comparison with measurement,
Radio Sci., 20 (2): 229246, 1985.
75. R. A. Pappert Effects of a large patch of sporadic-E on night-time propagation at lower ELF, J. Atm. Terrestrial Phys.,
42, 417425, 1980.
76. R. A. Pappert W. F. Moler A theoretical study of ELF normal mode reflection and absorption produced by night-time
ionospheres, J. Atm. Terrestrial Phys., 40: 10311045, 1978.
20
77. J. R. Katan P. R. Bannister Summary of ELF propagation variations at mid and high latitudes during the November/December 1982 and February 1984 solar proton events, Radio Sci., 22 (1): 111124, 1987.
78. E. C. Field C. R. Warber R. G. Joiner Effects of the ionosphere on ELF signals during polar cap absorption event:
Comparison of theory and experiments, in H. Soicher, (ed.), AGARD Conf. Proc. No. 382: Propagation Effects on
Military Systems in the High Latitude Region, North Atlantic Treaty Organization, 1985, 8.3-1-8.3-10.
79. A. C. Fraser-Smith R. A. Helliwell ELF sferic occurrences in the antarctic during a solar proton event: Case study of
occurrences at Byrd station during the event of June 9, 1968, J. Geophys. Res., 85 (A5): 22962306, 1980.
80. C. Greifinger P. S. Greifinger Noniterative procedure for calculating ELF mode constants in the anisotropic earth
ionosphere waveguide, Radio Sci., 21 (6): 981990, 1986.
81. R. A. Pappert W. F. Moler Propagation theory and calculations at lower extremely low frequencies (ELF), IEEE Trans.
Commun., COM-22 (4): 438451, 1974.
82. E. C. Field Propagation of ELF waves under normal and naturally disturbed conditions, J. Geophys. Res., 74 (14):
36393650, 1969.
83. V. I. Aksenov M. V. Nazarova Numerical solution of the problem of transmission of ELF waves through the lower
ionosphere, Radio Eng. Electron. Phys., 17 (7): 10571064, 1972.
84. J. Galejs On the terrestrial propagation of ELF and VLF waves in the presence of a radial magnetic field, Radio Sci. J.
Res. NBS/USNC-URSI, 69D (5): 705720, 1965.
85. C. Altman H. Cory The simple thin-film optical method in electromagnetic wave propagation, Radio Sci., 4 (5): 449457,
1969.
86. C. Altman H. Cory The generalized thin-film optical method in electromagnetic wave propagation, Radio Sci., 4 (5):
459470, 1969.
87. I. Nagano M. Mambo G. Hutatsuishi Numerical calculation of electromagnetic waves in an anisotropic multilayered
medium, Radio Sci., 10 (6): 611617, 1975.
88. E. C. Field R. G. Joiner An integral-equation approach to long-wave propagation in a nonstratified earthionosphere
waveguide, Radio Sci., 14 (6): 10571068, 1979.
89. J. A. Ferguson L. R. Hitney R. A. Pappert A program to compute vertical electric ELF fields in a laterally inhomogeneous
earthionosphere waveguide, Technical Report NOCS TR 851, Naval Ocean Systems Center, San Diego, CA, 1982.
90. C. H. Shellman A model for propagation of ELF waves throughout the lateral extent of the inhomogeneous earth
ionosphere waveguide, Radio Sci., 24 (1): 3546, 1989.
DANA PORRAT
ANTONY C. FRASER-SMITH
Stanford University
319
d0
100
t ( s)
(a)
d = 1381 km
Re E( ,d) (v/m)
d = 1065 km
77 100
j=0
248
46
j=0
100
j=1
j=1
j=2
j=2
t ( s)
t ( s)
(b)
(c)
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
320
us use some simple intuition and a simple model for the propagation environment. Consider the relationship between time
and frequency:
E (t , d) =
1
2
exp(it) E (, d) f s () d
[1,1]
Tee
(1)
j=1
[2,1]
Tem
[2,2]
TmeTee
Tee
j=2
Rm Re
[2,1]
(a)
[3,2]
[3,1]
[3,3]
Tme Tee T
Tme Tem
Tmm Tee
Tem
Tee
me T ee
j=3
Rm Re Rm Re
[3,1] Rm Re
[4,1]
Tem
Tee
[4,2]
Tme Tem
Tmm Tee
[3,2]
[4,3]
[4,4]
Tme Tem Tme Tem
Tee T
Tmm Tee Tmm Tee
ee
Tme Tme Tee Tee
Tme T
me
j=4
Rm Re
[4,1]
Rm Re Rm Re
[4,2]
Rm Rm R R
m m
Re Re R R
e e
[4,3]
(b)
TM
TE
Figure 2. (a) Diagrammatic representation of LF sky waves, depicting an indexing method for identification of a series of time-domain pulses traveling laterally between the ground and the ionosphere, j 1, 2, 3, . . .. Reflecting regions at the ionosphere are
indicated as ( j, k), j 1, 2, 3, . . ., k 1, 2, 3, . . .. (From Ref. 21)
(b) Diagrammatic representation of LF sky waves, depicting an indexing method for identification of a series of time-domain pulses
traveling laterally between the ground and the ionosphere, j 2, 4,
. . .. Reflecting regions at the ionosphere are indicated as ( j, k), j
1, 2, 3, . . ., k 1, 2, 3, . . .. (From Ref. 21.)
where
Rj1
( j 1)th earth reflection coefficient
e
p0j ionosphereground curvature focusingdefocusing
factor.
If the ionosphere is simply imperfectly reflecting with a finite
conductivity, an ionospheric reflection coefficient T can be introduced:
C j = (Tee ) j p0j Rej1
C1 = Tee
2
C2 = Re Tee
+ Rm Tem Tme
..
.
0
Re
Ge =
0 1
Tee Tem
T=
Tme Tmm
1
0
1 0
,
1=
Gm = p 0
0 Rm
0 1
xj
j1
j Cj
(Gm TGe ) Gm T = p0
yj zj
(5)
(6)
(7)
(8)
(3)
3
C3 = 2Re Rm Tee Tem Tme + R2e Tee
+ R2m Tmm Tem Tme
321
THEORY OF PROPAGATION OF
LOW-FREQUENCY SKY WAVES
Since Eq. (1) is independent of Maxwells equations, it is necessary only to construct a model for the propagation of the E
field in the frequency domain (2,11). The propagation model
usually employed to describe the LF sky waves in their natural environment is a finitely conducting spherical earth with
radius r a. This spherical earth is surrounded by a magnetoplasma, comprising electrons, ions, and neutral particles
with a superposed terrestrial magnetic field and a finite frequency of collision between particles. These particles go from
random into orbital motion when excited by the LF electromagnetic waves. Each particle has a finite collision frequency,
which tends to dampen the activity. Since the earths magnetic field changes the particle motion from linear to orbital,
and the direction of propagation of the EM wave may vary
with respect to the terrestrial magnetic field, the ionospheric
reflection process becomes anisotropic.
The detailed structure of the lower ionosphere between approximately 60 km and 100 km above the surface of the earth
is well documented in the literature (3,17). In the daytime, a
layer of such plasma placed at 60 km and concentric with the
earth usually serves as a simple model. This can be improved
by introducing a large number of such concentric shells of
plasma as a function of altitude, to take account of the variation of electron and ion density with altitude. Application of
Maxwells equations to this model results in a rigorous fullwave solution, if the ionosphere is assumed to be isotropic (1).
As is indicated in the discussion leading to Eq. (8) and in
Refs. 4, 1416, the curved, concentric, isotropic ionosphere reflection coefficient is normally replaced by a suitable planar
anisotropic reflection coefficient. This replacement involves
the magneto-ionic theory with a full-wave (plane-wave) reflection coefficient. The spherical-wave focusing effect of the
curve ionosphere can be retained in this process.
The LF radio wave field for a transmitter and receiver on
the surface of the earth separated by a distance d can be represented rigorously by a system of waves traveling in the radial (r) direction:
(4)
Er = A
n=0
G(, n)F(r, n)
(9)
322
Here
A=
I0 L0 c
4k21 a4
=
c
(p0 RTee )
j=1
(1) (2)
1a
1a
(13)
for
2 i 2
0
(14)
Er = Er,0 +
Er, j
(15)
j=1
2n + 1
1
z Pn (z)
P (z),
n+1
n + 1 n1
n = 1, 2, . . .
Er,0 = B
n=0
Finally,
(1) (2)
G(, n)1a
1a (1 + Re )
(16)
Here
(1) (2)
F (r, n) = 1a
1a
(1 + R)(1 + p0 Tee )
1 p0 RTee
B=
(10)
0 c I0 L
8 k21 a4
where
p0 =
(1) (2)
1a
1g
(2)
1a
1 + Re =
(11)
(1)
1g
2n + 1 (1,2)
(1,2)
n (z) n1
(z)
z
0(1,2) (z) = exp( iz)
(1,2)
n+1
(z) =
(12)
(1,2)
1
(z) = exp( iz)
(1,2)
(1,2)
means 1
(k1a). The earths reflection
The abbreviation 1a
coefficient for the spherical waves is
k1
ln 2a
k2
Re =
k
(2)
ln 1a
+ 1 ln 2a
k2
ln 1a
2i
(2)
(1) (2) [ ln (2) + (k /k
1a
1
2 ) ln 1a ]
1a
1a
n(1,2) (z) =
ln 1a
n (z)
=
n (z)
,
z=k 1 a
d
n (z)
n (z) =
dz
d (1,2)
dz n
The quantity B contains I0L, the source dipole current moment, and I0L 1 A m determines the E-field amplitude.
A particular sky wave can now be written
Er, j = B
n=0
z=k 1 a
(1) (2)
G(, n)1a
1a (1 + Re )2 p0 Rej1 Teej
(17)
The anisotropic sky waves can now be written using Eqs. (5)
(8):
F(r, n) = (1 + Re )
|1 + Gm T| (1) (2)
|1 Gm Ge T| 1a 1a
(18)
(1) (2)
F (r, n) = 1a
1a (1 + Re )I + (I + Ge )
(Gm Ge T ) j1 Gm T
j=1
(19)
with
I=
1
0
0
1
and so
Er, j =
n=0
(1) (2)
G(, n)1a
1a (1
+ Re )
p0j C j
!j
|I + Gm T|
= I +
Gm Ge T (I + Gm T )
|I Gm Ge T|
j=1
Here
3
(1,2)
v1/2
(z) =
z
2
(1)
1a
where
D2a =
(m)
2a
(m)
2a
(1)
s
Tee
(25)
k1
D (2)
k3 3g 1g
s
Tee =
k
(1) + 1 D (1)
1g
k3 3g 1g
(26)
v
,
k1 g
j = sin1
v
k1 a
(27)
Using the expansion given in Eq. (21) for an anisotropic ionosphere and integrating each term of the series along a suitable contour in the complex v plane again gives the ground
wave analog of Eq. (16) and a series of terms, each of which
is a particular sky wave analog of Eq. (17).
The reduction of Eq. (22) is detailed in Refs. 13, 19.
The wave hop series now becomes
Er =
f (v)(1 + Re ) dv
c
aj
cj
j=1
bj
a j1
= G1m Ge T
dj
c j1
b j1
d j1
(28)
(29)
where
Hv(1,2) (z)
k1
D (1)
k2 2a 1a
Re =
k
(2) + 1 D
1a
(2)
k2 2a 1a
(2)
1g
(23)
(2)
1g
(2)
1g
(21)
v
(1) (2)
P
( cos )1a
1a
cos v v1/2
s
T = Tee
=
(20)
f (v) = iA
323
(24)
Gm
p0
a1 = 1
G1m =
c1 = 0
a j = Re (Tee a j1 + TmeC j1 )
for
j2
c j = Rm (Tem a j1 + Tmm c j1 )
for
j2
Notwithstanding the fact that only TM waves have been excited and only TM waves are received at the receiver, a TEtype ground reflection coefficient is required at the ground
as a result of the effects of the earths magnetic field on the
ionosphere. Thus, due to anisotropy, the TE ground reflection
coefficient is required, as found explicitly in Refs. 19, 23:
k2
D (1)
(2)
k1 2a 1a 1a
Rm =
(1)
k
(2) 1a
(2) + 2 D
1a
2a 1a
k1
v
k2 a
2
(1)
1a
(30)
324
a L sin i cos a,
a 0 S 22
a1 2
hLhTaL
S2
s(s 2 h 2 )
a2 2 1
a3 2
X
Figure 3. Coordinate system for each reflecting region of the ionosphere.
Tee ,
Tmm ,
Tem ,
h 2L
h 2L a 2T
s
s2
S2 2
2
2
2
s h
s(s h )
s(s h 2 ) s(s 2 h 2 )
2
s 2 h 2L
s(s 2 h 2 )
1i ,
2N
H
,
2N
S 2 sin 2 i 1
h L h sin I,
h T h cos I
2N
N e2
0m
0 e Hm
gyrofrequency
m
hLhTaL
a L sec 2 i
s(s 2 h 2 )
a4 1
Hm
a T sin i sin
a 21 h 2T
s2
s
s1
1
2
1 2
S2
2
2
s h
s s h
s(s 2 h 2 )
s(s 2 h 2 )
N and m are the electron number density and mass. H m is the earths magnetic
intensity; I is the madnetic declination.
Tme
The details leading to the mathematical theory of this reflection process are given in Refs. 4, 14, 15, and 16. Consider one
of the reflecting regions at the ionosphere depicted in Fig.
Coupling
oe
eo
Zp=
Waves
Np, p p
Zp-1
Zn+1
Zn
Nn, n n
Z6
N6, 6 6
ir
ri
Z5
Upgoing (i)
Downgoing (r)
Coupling N5, 5 5
Z4
N4, 4 4
Z3
N3, 3 3
Z2
N2, 2 2
Z1
N1, 1 1
0 = 1
i
Hm
325
Tem
Tmm
aoe
aom
Tee
Tme
boe
bom
(1)
U eio
(1)
U mio
coe
com
(1)
eic
(1)
mie
doe
dom
(1)
U ero
(1)
U mro
(1)
U ere
(1)
U mre
(2)
eio
(2)
U mio
(2)
eie
(2)
U mie
(2)
(2)
U ero
U mro
(2)
U ere
(p1)
eio
(p1)
U mio
(p1)
eie
(p1)
U mie
ap(p4) ap(p9)
(p1)
(p1)
U ero
U mro
bp(p4) bp(p9)
(p1)
(p1)
U ere
U mre
cp(p4) cp(p9)
U (p)
U (p)
eio
mio
dp(p4) dp(p9)
U (p)
U (p)
mie
eie
a 11 cos i
a 12 cos i sin a
(1)
a 13 Q 1o
b 11 sin i
b 12 cos i cos a
b 13 1
a 14 Q ie(1)
(1)
a 15 Q ro
(1)
a 16 Q ro
b 14 1
b 15 1
b 16 1
z 1 io(1)
c
a 24 a 14 exp i z 1 ie(1)
c
(1)
a 25 a 15 exp i z 1 ro
c
(1)
a 26 a 16 exp i z 1 re
c
a 27 Q io(2)
a 28 Q ie(2)
(2)
a 29 Q ro
a 37 a 27 exp i z 2 io(2)
c
.
..
a p( p9) Q 1e
(2)
U mre
a 23 a 13 exp i
z 1 io(1)
c
b 24 exp i z 1 ie(1)
c
(1)
b 25 exp i z 1 ro
c
(1)
b 26 exp i z 1 re
c
b 27 1
b 28 1
b 23 exp 1
b 29 1
b 37 exp i
.
..
b p( p9) 1
z 2 io(2)
c
c 11 cos i sin a
c 12 cos a
c 13 (a L P io(1) io(1))
c 14 (a L P ie(1) ie(1))
(1)
(1)
ro
)
c 15 (a L P ro
d 11 cos i cos a
d 12 sin a
d 13 ( io(1) Q io(1) a T P io(1))
d 14 ( ie(1) Q ie(1) a T P ie(1))
(1) (1)
(1)
d 15 ( ro
Q ro a T P ro
)
(1)
(1)
re
)
c 16 (aL P re
(1) (1)
(1)
d 16 ( re
Q re a T P re
)
c 23 c 13 exp
c 24 c 14 exp
c 25 c 15 exp
c 26 c 16 exp
i z 1 io(1)
c
i z 1 ie(1)
c
(1)
i z 1 ro
c
(1)
i z 1 re
c
c 27 (a L P 2io io(2))
c 28 (a L P ie(2) ie(2))
(2)
(2)
c 29 (a L P ro
ro
)
(2)
(2)
)
c 2 10 (a L P re re
c 37 c 27 exp i z 2 io(2)
.
c
..
c p( p9) (a L P (iep) (iep))
z 1 io(1)
c
d 24 d 14 exp i z 1 ie(1)
c
(1)
d 25 d 15 exp i z 1 ro
c
(1)
d 26 d 16 exp i z 1 re
c
d 23 d 13 exp i
(2)
d 37
. d 27 exp i c z 2 io
..
d p( p9) ( (iep) Q (iep) a T P (iep))
326
a oe cos i sin a
a om cos a
b oe cos i sin a
b om sin a
c oe cos a
c om cos i sin a
d oe sin a
d om cos i cos a
hT hL
s
A1 aL
1 a 2L 2 2
s(s 2 h 2 )
s h2
hL
hT
B 1 a L a T i 2
aT i 2
s h2
s h2
hL
s 2 h 2L
C 1 1 a 2s
aL aT i 2
s(s 2 h 2 )
s h2
hT
hLhT
D1 aT i 2
aL
s h2
s(s 2 h 2 )
s 2 h 2L
s
E 1 1 a 2s
1 a 2L 2 2
s(s 2 h 2 )
s h2
hT
hT
F1 aT i 2
aT i 2
s h2
s h2
A1 B1
C1 D1
P
Q
E1 F1
E1 F1
H
H
=0
t
E
E
H J 0
=0
t
E + 0
Here (n), P P (n), Q Q (n) for a particular slab (Fig. 4), and a s sin i ;
io, ie, ro, re refer to the four roots of Eq. (35).
wavefronts are contained in the xy plane. The directions relative to the magnetic field vector are as follows:
(31)
V
dV
V + 0 eV
V H + eE
E=0
+ mV
dt
(32)
(33)
107
Amplitude (V/m)
108
3
4
5
6
7
j=8
109
3
1010
5
4
Figure 5. Amplitude of the total E field as a function of distance along the surface of the earth, together with the individual wave hops, j 0 (the ground wave), 1, 2, 3, . . ., illustrating propagation in the presence of an ionosphere with infinite
conductivity. (From Ref. 21 by permission IEEE (2).)
j=0
j=1
6
j=2
7
8
1000 2000 3000 4000 5000 6000 7000 8000 9000 10,000
Distance (km)
1011
327
such as Refs. 21, 22. The complex index of refraction of a particular layer in the model now can be found:
E t | exp i t D
Et = |E
c
2 = 2 + sin2 i
(34)
(35)
120
Tee =
140
Tme =
160
U (n)eio =
4
5
Decibels
180
U (n)eie =
6
3
7
200
Ey r
Ey i
Ey r
Ex i
(n)
Eyio
Ey i
(n)
Eyie
Ey i
Tem =
Ex r
Ey
i
E
= xr
Ex i
Tmm
U (n)mio =
U (n)mie =
(n)
Eyio
Ex i
(n)
Eyie
Ex r
8
2
9
220
(36)
10
240
11
1
260
j =
280
300
2000
4000
6000
8000
10,000
Distance (km)
v
k1 a
328
107
j=0
Amplitude (V/m)
108
j=3
109
j=4
j=5
j=6
1010
j=2
j=7
j=1
j=8
j=0
Figure 7. Sample wave hop calculation using the full magneto-ionic theory for the model given in Table 2, for a propagation path over sea water with magnetic parameters and ionosphere electron density profiles given in Ref. 21.
1011
1000 2000 3000 4000 5000 6000 7000 8000 9000 10,000
Distance (km)
BIBLIOGRAPHY
1. J. R. Johler, Spherical wave theory for MF, LF, and VLF propagation, Radio Sci., 5: 14291443, 1970.
2. J. R. Johler, Propagation of the low-frequency radio signal, Proc.
IRE, 50 (4): 404427, 1962.
3. J. S. Belrose, L. A. Bound, and L. W. Hewitt, Ground based radio
wave propagation studies of the lower ionosphere, Proc. Conf.,
held 1115 April 1966, Defence Research Telecommunications
Establishment, Radio Physics Lab., Defense Research Board,
Dept. of National Defence, Shirley Bay, Ottawa, Canada, 1967.
4. J. R. Johler and L. A. Berry, On the effect of heavy ions on LF
propagation, with special reference to a nuclear environment,
Natl. Bur. Stand. Tech. Note 313, Washington: U.S. Govt. Printing Office, 1965.
5. D. R. Hartree, Proc. Cambridge Philos. Soc., 25: 27, 1927.
6. D. R. Hartree, The propagation of electro-magnetic waves in a
refracting medium in a magnetic field, Proc. Cambridge Philos.
Soc., 27: 143, 1931.
7. E. V. Appleton, J. Inst. Elec. Eng., 71: 642, 1932.
8. J. A. Ratcliffe, The Magneto-ionic Theory and Its Applications to
the Ionosphere, Cambridge, UK: Cambridge Univ. Press, 1959.
9. J. R. Wait, Electromagnetic Waves in Stratified Media, New York:
Pergamon, 1962.
10. G. Hefley, The Development of Loran-C Navigation and Timing,
NBS Monograph 129, Washington: U.S. Govt. Printing Office,
1972.
11. J. R. Johler, Propagation of an electromagnetic pulse from a nuclear burst, IEEE Trans. Antennas Propag., AP-15: 256263,
1967.
12. H. Bremmer, Terrestrial Radio Waves, New York: Elsevier, 1949.
13. L. A. Berry, Wave hop theory of long distance propagation of LF
radio waves, J. Res. Natl. Bur. Stand., 68D: 12751282, 1964.
329
J. RALPH JOHLER
Johler Associates
pinging on the ground. Also, in this frequency range atmospheric effects are small. Thus the Friis formulation is directly applicable. In other situations the free-space equation
may be the starting formulation to which corrections for other
effects (e.g., atmospheric attenuation, attenuation due to the
ground) can be added. In fact, attenuation over a path is often
specified as the attenuation relative to that for the same distance in free space; the latter would be calculated by the
Friis formula.
(1)
Far Field
In this expression Ar denotes the effective area of the receiving antenna in the direction of the transmitter, At is the effective area of the transmitting antenna in the direction of the
receiver, is the wavelength, and R is the distance separating
the transmitting and receiving antenna, all measured in compatible units (e.g., both powers in watts, both areas in square
meters, and R in meters). An alternative form is
G Gr 2
Pr
= t 2 2
Pt
(4 ) R
(2)
where Gt and Gr denote, respectively, the gain of the transmitting antenna in the direction of the receiver and the gain of
the receiving antenna in the direction of the transmitter, and
and R must be given in the same units. For convenience, a
decibel formulation is sometimes used. Such formulations are
obtained by setting c/f, where c is the free-space velocity
of light and f the frequency, converting quantities to the desired units (e.g., R from meters to kilometers) taking the common logarithm of both sides of the resulting equation, and
multiplying by 10. Algebraic and arithmetic manipulation
then yield expressions such as
Pr,dbW = Pt,dbW + Gt,dB + Gr,dB 20log10 Rkm
20log10 f MHz 32.4
Polarization
In this formulation it has been assumed that the polarization
of the wave at the receiving antenna is optimal for that antenna; otherwise a polarization-mismatch factor (in the decibel formula, a polarization-mismatch term) must be included.
DERIVATION
Equations (1) and (2) can be derived easily from basic physical principles. For a hypothetical lossless, isotropic transmitting antenna the power applied to the input terminals would
be spread uniformly over a sphere at radius R, giving a flux
density
(3)
Si =
Pt
4R2
(4)
Pt Gt
4R2
(5)
ASSUMPTIONS
Free Space
Pr = SAr =
Pt Gt Ar
4R2
(6)
4A
2
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
(7)
HISTORICAL NOTE
The formulas in Eqs. (1) and (2) are named after Harald Trap
Friis, who published Eq. (1) in 1946 while he was a research
engineer at Bell Telephone Laboratories, Inc. (1). Mr. Friis
later became Director of Research in High Frequency and
Electronics at Bell Telephone Laboratories and was honored
with numerous medals and awards for his technical work and
his leadership.
BIBLIOGRAPHY
1. H. T. Friis, A note on a simple transmission formula, Proc. IRE
(subsequently Proc. IEEE ) 34: 254256, 1946.
Reading List
An interesting summary of derivations of the Friis formula from various perspectives can be found in D. C. Hogg, Fun with the Friis freespace transmission formula, IEEE Antennas Propag., Mag. 35 (4): 33
35, 1993.
CURT A. LEVIS
The Ohio State University